docs(index): mxaccessgw Contracts/Client 0.3.0 published; main e5e7f18

docs(index): HistorianGateway main now gRPC-only (swap-historian-client merged, 99b4a7d)
docs(index): refresh ScadaBridge/HistorianGateway rows; record HG 2026-10-16 diagnostics + Int1 milestones
2026-11-06 16:01:22 -05:00 · 2026-11-06 15:52:22 -05:00 · 2026-11-06 15:38:26 -05:00 · 2026-11-06 15:36:33 -05:00 · 2026-11-06 15:34:30 -05:00 · 2026-11-04 00:52:25 -05:00
299 changed files with 32749 additions and 1234 deletions
@@ -0,0 +1,16 @@
+---
+name: sister-project-history
+description: Use when you need the detailed change history, live-gate records, redundancy/auth/ACL behaviour, or ScadaBridge consumer-impact notes for a sister project (OtOpcUa, HistorianGateway, BTDB) — the scadaproj root CLAUDE.md keeps only a short summary row per project; the full narrative lives in this directory.
+---
+
+# Sister project history
+
+The root `CLAUDE.md` index keeps one short row per sister project. The full narrative rows that used to live
+in that table were moved here **verbatim** on 2026-09-03 (they were ~57k characters loading into every session):
+
+- `otopcua.md` — OtOpcUa: v3 dual-namespace address space, every driver, redundancy/mesh program, data-plane auth + ACL enforcement, deferment register, the authoring-vs-deploy-gate parity arc, and the per-issue consumer-impact record for ScadaBridge.
+- `historiangateway.md` — HistorianGateway: ArchReview cycles, pending.md items, session pools, datatype gaps, SQL-path workarounds, PR/commit history.
+- `btdb.md` — BTDB (Batch Tracking Database, added 2026-09-03): schema + sproc write contract, Historian/CLR coupling, Galaxy/MESAPI/ScadaBridge edges, the DARS `BatchTracker_DB` fork, and the gap analysis.
+
+Read the file for the project you are working against. When a sister repo tells you to "update its entry in
+`scadaproj/CLAUDE.md`", append the detail to the file here and keep the index row to a few lines.
@@ -0,0 +1,207 @@
+# BTDB (Batch Tracking Database) — index history + analysis
+
+> Added 2026-09-03 as a sister project. **This file is the authoritative place for BTDB detail — append here and keep the root index row short.** Everything below was verified from source on 2026-09-03 unless marked _(inferred)_.
+
+| Project | Location | Stack | Repo | Summary |
+|---|---|---|---|---|
+| **BTDB** (Batch Tracker) | `~/Desktop/btdb` (case-insensitive FS: also `BTDB`) | SQL Server 2016+ SSDT project + SQL CLR (net472) · manager: **Blazor Web App net10.0** on `feat/manager-blazor` (§8; `master` still carries the EOL net6.0 MVC app) | **`ZimmerBiomet/BTDB`** on the corporate Gitea `wonder-app-vd03.zmr.zimmer.com:3000` (`master` @ `d147c10e`, 2023-07-26). **Not** on `gitea.dohertylan.com`. | The `BT` SQL Server database that is the plant's **batch/cycle/work-order system of record for data collection**: Galaxy machine objects call its `StartMachineBatch`/`StartMachineCycle`/`End*`/`AddMachineBatchWorkOrder`/`AddOtherData*` sprocs; reports pull time-series per batch out of the AVEVA Historian through its CLR sprocs (`OPENQUERY(INSQL, … WideHistory …)` + `Runtime.dbo.*`); MESAPI and ScadaBridge's inbound API resolve `SAPID → Machine.Code` from `dbo.Machine`. `BTDBManager` is a small Windows-service MVC site for machine/group config, batch edits (audited) and cross-environment config import. |
+
+## 1. What is on disk, and where the truth is
+
+- `~/Desktop/btdb` is **not a git checkout** — no `.git`, no `CLAUDE.md`, no `README`. It is a file-for-file snapshot of `ZimmerBiomet/BTDB` `master` (the only branch) at its head commit `d147c10e` (2023-07-26, "Update notification system tables to add urgent flag"). Verified by comparing the Gitea tree with the local tree: 170 vs 168 files, the only difference being the remote's `.gitattributes`/`.gitignore`. The repo's `updated_at` of 2024-11-07 is not a commit (the branch head is still 2023-07-26).
+- `~/Desktop/DARS/WW_ARCH/BatchTracker_DB/` is a **fork of the `Database` project only** (no Manager), modified 2026-08-13, also not under git (DARS is not a repo). Substantive differences vs the snapshot, all matching the DARS sprint-1 spec `DARS/sprints/sprint1/bt_db_changes.md`:
+  - new table `MachineBatchTransfer` (source/dest batch chain for mid-run reactor failure; `TransferType` CHECK `'mid_run_failure'`; three indexes) + sproc `AddMachineBatchTransfer` (linear-chain guard);
+  - `MachineBatchWO` gains `ContainerID NVARCHAR(64) NULL` + `MoveOutTimestamp DATETIME2 NULL` with filtered indexes (Camstar container traceability; mirrors the `SAPID`/`ZTag` idiom);
+  - new CLR sproc `GetLinkedBatchTSV` (+ `_Wrapper`) that walks the transfer chain and stitches TSVs, and `LinkedBatchAlarm_Wrapper`;
+  - `StartMachineBatch.sql`, `MachineBatch.sql`, `AuditLogMachineBatch.sql` differ **only** in BOM/CRLF → LF normalization (no logic change).
+  - Reference doc for the schema: `DARS/WW_ARCH/BatchTracker_Reference.md` (states 22 tables / 35 SQL sprocs / 4 CLR sprocs / 5 CLR UDFs / 6 table types / 1 aggregate).
+- **Resolved 2026-09-03 — the divergence is closed locally:** `~/Desktop/btdb` was turned into a real checkout (`git init` + fetch of `ZimmerBiomet/BTDB`, `master` tracking `origin/master`; the snapshot was tree-identical apart from exec bits copied from Windows, which were cleared). The DARS fork's transfer-chain work was migrated onto branch **`feat/machine-batch-transfer`** as one commit: the five new objects, the two `MachineBatchWO` columns, and the `.sqlproj` wiring **including the `Compile` entry for `GetLinkedBatchTSV.cs` that the fork had omitted** (without it the CLR sproc is never built and its wrapper fails at deploy). The BOM/CRLF re-saves were not carried over. The sprint-1 plan (`bt_db_changes.md`, overview, diagrams) and `BatchTracker_Reference.md` now live under `btdb/docs/`. **The branch is local only — not pushed to the corporate Gitea.** `DARS/WW_ARCH/BatchTracker_DB/` is now superseded and should be deleted or replaced by a pointer once the branch is pushed. A second commit on the same branch carried the two columns through the Manager's batch-edit path: `MachineBatchWOType` (appended last — the TVP binds by ordinal and `ToDataTable` emits properties in declaration order, so the view model's property order must keep matching), `AuditLogMachineBatchWO` (`Prev_`/current pairs), the DTOs and view model, `SaveBatchEdit.sql` (MERGE change-detection with `COALESCE` because both columns are NULL-able, OUTPUT and the three audit inserts), and the Kendo work-order grid in `Views/BatchEdit/Edit.cshtml` (masked timestamp editor like the cycle grid). `LookupBatchInfo.sql` is `SELECT *` so it needed nothing. Not touched: the Telerik `BatchEditAuditLogReport.trdp` (binary package; open it in the Telerik designer to add the audit columns) and the batch search grid (`SearchBatches.sql`), which still lists only work-order numbers. Nothing here was built or run: the Manager needs Telerik + Windows, and the SSDT project needs MSBuild; a throwaway compile of the Manager retargeted to .NET 10 on the Mac showed no errors in the touched files.
+
+## 2. Solution shape
+
+`BTDB.sln` (VS 2022) = two projects:
+
+| Project | Kind | Notes |
+|---|---|---|
+| `Database/Database.sqlproj` | SSDT, `Sql130DatabaseSchemaProvider` (SQL Server 2016+), CLR assembly targets **net472**, C# namespace `BT` | 22 tables, 6 table types, ~35 T-SQL sprocs, 3 CLR sprocs (`GetTSV`, `GetMachineBatchTSV`, `GetCycleTSV`), 5 CLR UDFs (`GetAggregateValues`, `CheckTriggered`, `WhenTriggered`, `WhenCleared`, `StringCompare`), 1 CLR aggregate (`StringJoin`), 1 T-SQL TVF (`GenerateShifts`), 1 trigger (`TrgNotification` → `AuditLogNotification`). Requires `CLR enabled` on the instance. |
+| `Manager/BTDBManager.csproj` (**`master` only — deleted on `feat/manager-blazor`, see §8**) | ASP.NET Core MVC, **net6.0** (out of support since 2024-11), Dapper 2.0, `System.Data.SqlClient`, `System.DirectoryServices` (Windows-only LDAP), `Microsoft.Extensions.Hosting.WindowsServices`, Kendo UI + Bootstrap, **Telerik Reporting R2 2022** referenced by an absolute `HintPath` into `Program Files (x86)` | Kestrel `http://*:9876`; runs as a Windows service; 13 embedded `.sql` query resources; 3 `.trdp` Telerik audit reports. `Manager/install.bat`/`uninstall.bat` are **leftovers from a different service** (`WWNotificationSystem.exe`) and do not install the Manager. |
+
+## 3. Database model (what the family writes and reads)
+
+**Core hierarchy:** `Machine` (`Code` unique; optional unique `ZTag`, `SAPID`; `TimeZone`; `MultipleBatch`/`MultipleCycle` flags; `OtherData` JSON) → `MachineBatch` (start/end operator + UTC timestamps, `OtherData`) → `MachineCycle` (same shape) and `MachineBatchWO` (`WorkOrderNumber`, `ItemNumber`). `MachineGroup` + `MachineGroupMembership` give roll-ups (DARS models one `Machine` per reactor *side*, one `MachineGroup` per unit). `MachineAlarm` is a per-machine alarm catalog (`Severity` 0–999, `FlaggedForMES`).
+
+**Lifecycle sprocs (the write contract):**
+
+| Sproc | Behaviour |
+|---|---|
+| `StartMachineBatch(@Code, @OperatorID, @o_MachineBatchID OUT)` | Looks up machine by `Code`; unless `MultipleCycle`/`MultipleBatch`, **auto-closes** any open cycle/batch first; inserts with `GETUTCDATE()`. Throws 50000 if machine unknown. |
+| `EndMachineBatch(@MachineBatchID, @EndOperatorID)` | Closes the open cycle too; throws if not found / already closed. |
+| `StartMachineCycle(@MachineBatchID, @OperatorID, @o_MachineCycleID OUT)` / `EndMachineCycle` | Cycle must belong to an open batch. |
+| `AddMachineBatchWorkOrder(@MachineBatchID, @WorkOrderNumber, @ItemNumber, @o OUT)` | Only on open batches. |
+| `AddOtherDataMachineBatch` / `…MachineBatchWorkOrder` / `…MachineCycle (@ID, @DataKey, @DataValue)` | `JSON_MODIFY` into `OtherData` — via **dynamic SQL that concatenates `@DataKey`/`@DataValue`** into the statement (`sp_executesql` of a built string). Injection surface; callers today are trusted Galaxy scripts. |
+
+All timestamps are `GETUTCDATE()`; `Machine.TimeZone` is used only at retrieval time (`wwTimeZone=` in report wrappers). This matches the family's all-UTC rule.
+
+**Historian coupling (the read contract):** the CLR code resolves a machine-relative tag (`<Code>.<local>`; `*.<attr>` shorthand; `|literal|` static columns) to the Historian's global tagname via `Runtime.dbo.TagExtendedPropertyInfo` (`HierarchicalName`), then runs `SELECT … FROM OPENQUERY(INSQL, 'SELECT DateTime, [tag]… FROM WideHistory WHERE … wwTimeZone="UTC" AND <queryParams>')` (`wwRetrievalMode`, `wwResolution`, `wwTimeZone` are passed through **verbatim** from `@queryParams`). Discrete checks read `Runtime.dbo.DiscreteHistory`; alarm wrappers read `Runtime.dbo.Events` (`IsAlarm=1`, `Alarm_OriginationTime`, `Alarm_DurationMs`, `Alarm_UnAckDurationMs`). **Consequence: the `BT` database must live on (or three-part-name-reach) the Historian's SQL instance with the `INSQL` linked server** — it is co-tenant with `Runtime`, which is why it sits on the `wonder-sql-*` boxes.
+
+**Report wrappers** are per machine family and are the consumers of the CLR layer: Crest (`CrestWide`/`CrestMinMax`/`CrestAlarm`/`CrestCleaner*`), Elma (`ElmaData`, `Elma_Bulk_Query_Nitric`), `ExtruderWide`, `PultrusionWide`, Haas (`HaasStatusReport`, `HaasAlarm`, `QV_Haas*` = QlikView), Solar (`SolarAlarms`/`SolarEvents`), DryHeat (`DryHeatCycleReport`/`Search`/`BulkReport`), PoncePolyPress, `Z002432Wide`/`Z002433Wide`. `GenerateShifts` builds a 3-shift calendar table for these.
+
+**Auth tables that are not the Manager's:** `UserAuth`, `UserAuthDetails`, `UserAuthRole`, `ApiKey` are ServiceStack's schema — they belong to **MESAPI/WWSupport** (the Camstar-facing REST service, see `mes-delmia-integration-api.md` §1.1: `ApiKeyAuthProvider` + `LdapAuthProvider`, role `MESAPI`), which shares the `BT` database. The Manager authenticates via LDAP only.
+
+**Notification tables** (`Notification`, `NotificationLog`, audited by trigger) back the separate `WWNotificationSystem` Windows service, which is **not in this repo**.
+
+## 4. BTDBManager (the config/edit UI) — `master` state; the replacement is §8
+
+- **Auth:** cookie scheme; `LdapAuthenticationService` binds `LDAP://nam.zimmer.com` with the user's own credentials via `System.DirectoryServices` (so it only runs on a domain-joined Windows host), reads `memberOf`, and keeps only the three configured group DNs. Policies `READONLY` / `MACHINECONFIG` / `BATCHEDIT` map to `ReadOnlySecurityGroup` / `MachineConfigSecurityGroup` / `BatchEditSecurityGroup`; the checked-in DEV config points all three at `UG_WW_APP_ADMIN_DEV`. A global `AuthorizeFilter` makes every controller (incl. `/Api/GetMachines`, `/Api/GetMachineGroups`) require login. This is exactly the AD-group→role problem the family's `ZB.MOM.WW.Auth` 0.3.0 + `ldap_mapping.md` already solve.
+- **Features:** Machine CRUD (with group memberships + alarm catalog via TVPs `MachineGroupMembershipType`/`MachineAlarmType`), MachineGroup CRUD, batch search/edit (`SaveBatchEdit` writes `AuditLogBatchEdit` + prev/new `AuditLogMachineBatch`/`…WO`/`…Cycle` rows), **MachineConfigImport** (copies machine/group config from another environment's `BT` — sources `QA` and `PROD` are hard-wired in `appsettings.json`), and Telerik `.trdp` audit reports.
+- **Secrets are checked in:** `appsettings.json` carries plaintext `wonderapp` SQL passwords for DEV/QA/PROD, and `install.bat` carries a domain service-account password. Treat all of them as compromised-by-repo; do not copy them into any other file or reply. This is the pattern `ZB.MOM.WW.Secrets` (`${secret:}`) exists to replace.
+- **Environment hosts as recorded in 2023:** DEV `wonder-sql-vd01/BT`, QA `wondersqlqa/BT` (`MultiSubnetFailover`), PROD `wondersqlpd/BT`. ScadaBridge's 2026-06-16 design says the DEV BTDB is on **`wonder-sql-vd03/BT`** (db-connection id 1). `mes-delmia-integration-api.md` also says `vd01`. **Host drift is unresolved** — do not assume either without probing.
+
+## 5. How BTDB relates to the rest of the family
+
+```
+   Galaxy machine objects ──(script: EXEC StartMachineBatch / StartMachineCycle /
+   ($Reactor_Side etc.)      AddMachineBatchWorkOrder / AddOtherData* …)──────────┐
+                                                                                  ▼
+   Camstar MES ──HTTPS──▶ MESAPI/WWSupport ──(SELECT Machine by SAPID; alarm catalog;  ┌────────────────────┐
+                          (ServiceStack, own UserAuth/ApiKey tables) ────────────────▶ │  SQL Server "BT"   │
+                                                                                      │  (BTDB)            │
+   ScadaBridge inbound API ──(Database helper, connection "BTDB": SELECT Code FROM     │  co-tenant with    │
+   (IpsenMESMoveIn, MES alarm-status) ─────────────────────────────────────────────▶ │  Historian Runtime │
+   ScadaBridge site scripts ──(Database.Connection / CachedWrite — customer scripts)─▶ │  + INSQL linked    │
+                                                                                      └─────────┬──────────┘
+   BTDBManager (config, batch edits, env import) ─────────────────────────────────────▶         │ CLR: OPENQUERY(INSQL) WideHistory,
+   SSRS / QlikView / Telerik reports ──(EXEC *_Wrapper, GetMachineBatchTSV …)──────────▶         │ Runtime.dbo.{TagExtendedPropertyInfo,
+                                                                                                ▼ DiscreteHistory, Events}
+                                                                                  AVEVA Historian (Runtime DB)
+```
+
+- **Galaxy → BT** is the primary write path (batches/cycles/work orders/OtherData JSON). DARS `model_info.md` §: "Existing per-machine BT sprocs (`StartMachineBatch`, `StartMachineCycle`, `GetCycleTSV`, `Alarm_Wrapper`, …) work unchanged" — the sprocs are the contract Galaxy scripts bind to.
+- **ScadaBridge → BT (verified):** the inbound-API script context has a `Database` helper bound to named db-connections; production defines connection **`BTDB`** (id 1), and the `IpsenMESMoveIn` (design 2026-06-16) and MES alarm-status (2026-06-30) methods do `SELECT Code FROM dbo.Machine WHERE SAPID=@s` before `Route.To(instance)`. Template/site scripts expose `Database.Connection(name)` + `Database.CachedWrite(...)`, so site scripts **can** write batches into BT — but **no checked-in ScadaBridge script calls a BT sproc**; any batch logging is customer-site script configuration, not platform code, and the docker rig does **not** seed a `BT` database (`infra/mssql/setup.sql` creates only `ScadaBridgeConfig` + `ScadaBridgeMachineData`; the `InboundDatabaseHelperTests` use a stand-in). ScadaBridge's inbound API is the designated successor of MESAPI's move-in/move-out surface, which is why it needs the same `SAPID → Code` lookup.
+- **HistorianGateway ↔ BT:** no direct coupling, but both read `Runtime.dbo.Events` (HG's SQL-path `ReadEvents`) and both need the Historian SQL instance. BT's CLR `GetTSV`/`GetAggregateValues` is a bespoke, in-database equivalent of what HG's `ReadRaw`/aggregate retrieval modes do over gRPC — a candidate for consolidation if the report layer is ever re-platformed.
+- **OtOpcUa / mxaccessgw:** no coupling.
+- **Break surface:** the sproc signatures + `Machine.Code`/`SAPID`/`ZTag` semantics (Galaxy scripts, MESAPI, ScadaBridge inbound methods all bind to them), the `OtherData` JSON key conventions (DARS sprint 1 defines ~14 new keys with single writers), and the `INSQL`/`Runtime` co-location assumption.
+
+## 6. Analysis — gaps and recommendations (2026-09-03)
+
+1. **Push the branch.** ~~Put both copies under git~~ — done locally 2026-09-03 (`feat/machine-batch-transfer`). Remaining: push it to `ZimmerBiomet/BTDB`, review/merge, decide whether BTDB also gets a mirror on `gitea.dohertylan.com` like the other sisters, then delete the superseded `DARS/WW_ARCH/BatchTracker_DB/` copy.
+2. **Rotate and externalize the checked-in credentials** (three SQL passwords, one domain account). _Externalised on `feat/manager-blazor` (`${secret:btdb/...}` refs, store outside the content root); rotation is a cut-over step (§8)._ The family standard is `ZB.MOM.WW.Secrets` `${secret:}` with an absolute `Secrets:SqlitePath` outside the content root.
+3. **Manager re-platform** — _done on `feat/manager-blazor`, see §8_: net6.0 is EOL; Telerik is pinned by an absolute machine path; `System.DirectoryServices` binds the user's own credentials on a domain-joined box. A .NET 10 rewrite on the family libs is mostly assembly: Auth 0.3.0 (LDAP + `UG_WW_*` group→role mapping, `FallbackServers`), Theme 0.4.1, Health 0.3.0 (`MapZbHealth`, so Overview can watch it), Telemetry, Configuration, Audit (the batch-edit audit is a natural `ZB.MOM.WW.Audit` writer), Secrets. Keep Dapper + the embedded queries; replace Telerik `.trdp` with server-side rendering only if the three audit reports are still used.
+4. **Injection surfaces in the SQL layer:** `AddOtherData*` build dynamic SQL from `@DataKey`/`@DataValue`; `GetTSV`/`GetAggregateValues` interpolate tag names and append `@queryParams` raw into `OPENQUERY`. Trusted callers today, but ScadaBridge scripts and the inbound API widen the caller set. Fix `AddOtherData*` with parameterised `JSON_MODIFY` (no dynamic SQL needed), and whitelist `queryParams` to the three `ww*` keys.
+5. **ServiceStack auth tables** (`UserAuth*`, `ApiKey`) are MESAPI's, not BT's domain. Once ScadaBridge's inbound API fully replaces MESAPI (Ipsen MoveIn is the first slice; ScadaBridge keys already live in the `zbauth` schema via `Auth.ApiKeys.SqlServer` 0.3.0), those tables are dead weight in the schema and should be dropped from the sqlproj rather than carried forward.
+6. **Host inventory drift:** 2023 config says DEV BT is on `wonder-sql-vd01`; ScadaBridge's 2026 design says `wonder-sql-vd03`. Probe and record the answer in `deployment.md` §2.1 (which currently has no BT row at all).
+7. **`install.bat`/`uninstall.bat` are wrong for this project** (they install `WWNotificationSystem`). _Deleted on `feat/manager-blazor`; replaced by `deploy/install-service.bat` (NSSM, service `ZB.MOM.WW.BTDB.Manager`)._
+8. **CLR + Historian co-location is a hard deployment constraint** the index did not previously record: BT cannot be moved to the app-tier SQL (`wonder-app-vd03,1433`, where ScadaBridge's ConfigDb lives) without breaking every report wrapper. Any consolidation plan must keep BT on the Historian instance or replace the CLR with HistorianGateway calls.
+
+## 7. Primary commands (Windows only)
+
+```powershell
+# from ~/Desktop/btdb (master)
+dotnet build Manager\BTDBManager.csproj                 # net6.0; needs Telerik Reporting R2 2022 at the HintPath
+msbuild Database\Database.sqlproj /p:Configuration=Release   # SSDT (VS 2022 + SSDT) → bin\Release\Database.dacpac
+SqlPackage /Action:Publish /SourceFile:Database\bin\Release\Database.dacpac /TargetServerName:<historian-sql> /TargetDatabaseName:BT
+# Instance prerequisites: CLR enabled, `INSQL` linked server, `Runtime` DB reachable by three-part name
+```
+
+```bash
+# feat/manager-blazor (any OS for the app; SSDT still Windows-only)
+./build.sh                       # restore + build (Release, warnings-as-errors) + test BTDB.slnx
+dotnet run --project src/ZB.MOM.WW.BTDB.Manager   # appsettings.Development.json → local glauth; needs a BT
+pwsh deploy/publish.ps1          # self-contained win-x64 → deploy/out/ (+ NSSM install scripts)
+```
+
+```bash
+# Local Docker rig (2026-09-04, any OS) — see §9
+docker/deploy-db.sh              # BT on scadabridge-mssql from Database/ (dacpac + SQL CLR); --recreate drops it first
+docker/deploy.sh                 # publish FDD on the host → zb-btdb-manager:local, up on :9877 (glauth login enabled)
+```
+
+On `master` nothing builds on macOS (SSDT/CLR need Windows MSBuild, the MVC Manager needs Telerik + `System.DirectoryServices`). On the branch the app and its tests build and run on macOS; only `Database.sqlproj` (kept alone in `BTDB.sln`) still needs Windows — but §9's side projects build the same sources into a dacpac and a CLR assembly anywhere, for the local rig.
+
+## 8. Blazor conversion — branch `feat/manager-blazor` (2026-09-04)
+
+Plan: `btdb/docs/plans/manager_blazor_conversion.md` (+ `.tasks.json` tracker beside it; mockups `docs/mockups/manager_blazor_mockups.html`). Executed 2026-09-04 in worktree `~/.config/superpowers/worktrees/BTDB/feat-manager-blazor`; **uncommitted and unpushed at the time of writing** — phases 1–3 (scaffold, core, auth, widgets, queries, all pages) plus the local half of phase 4 are done; the DEV-box half of phase 4 (parity run, secrets seed + rotation, NSSM side-by-side, port swap) is written up as a runbook and not yet performed.
+
+- **Shape on the branch:** `BTDB.slnx` = `src/ZB.MOM.WW.BTDB.Manager` (Blazor Web App, InteractiveServer, net10.0) + `src/ZB.MOM.WW.BTDB.Manager.Core` (Dapper data layer moved verbatim from the MVC app, hardened) + two xunit/bUnit test projects (334 tests, warnings-as-errors gate via root `Directory.Build.props`, `build.sh`/`build.ps1`). `BTDB.sln` keeps only `Database.sqlproj`. `Manager/` (MVC, Kendo, the three Telerik `.trdp` reports with embedded connection strings, the mis-named `install.bat`) is **deleted**; `/Api/GetMachines`+`GetMachineGroups` are gone (no external consumer — ScadaBridge/MESAPI read `dbo.Machine` by SQL).
+- **Family libs:** Theme 0.4.1, Health 0.3.0 (`MapZbHealth`; readiness = a `SELECT 1` against `BT` with a 3 s timeout, so `/health/ready` is 503 until the DB answers — register it in Overview), Telemetry + Serilog 0.1.0, Configuration 0.1.0 (`ConfigPreflight` requires `ConnectionStrings:BT` and `Btdb:Security:Ldap:Server`), Secrets 0.6.2 (`${secret:btdb/connectionstrings/{bt,qa,prod}}`, `${secret:btdb/ldap/serviceaccountpassword}`; SQLite store `C:\ProgramData\ZB.MOM.WW\btdb-manager\secrets.db`, KEK from `ZB_SECRETS_MASTER_KEY`), Auth.Ldap/AspNetCore 0.3.0 (service-account bind, `Btdb:Security:GroupToRole` → `CanonicalRole`; policies Viewer / BatchEdit (Operator+) / MachineConfig (Administrator); cookie `btdb.auth`, plain-HTTP cookie because the box has no certificate). `Microsoft.Data.SqlClient` 7.x with `Encrypt=False` forced by `BtConnectionStrings.Normalize` unless the string sets it.
+- **Behavioural deltas vs the MVC app worth knowing:** config import derives the group set from the selected machines' memberships (no separate group pick) and has a dry-run preview (`MachineConfigImportPreview.sql`: same MERGEs, `OUTPUT` into a table variable, always `ROLLBACK`); batch-edit timestamps are 7-fraction UTC strings validated on blur (no Kendo mask); the three audit reports are HTML tables with a print stylesheet instead of Telerik PDFs; `MachineAlarm` DTO property order/types were wrong in the MVC app (alarm import always threw) and are fixed against `MachineAlarmType`; every `Save*` runs in a transaction and `THROW 5001x` maps to `BtdbSaveException`.
+- **Cut-over material:** `docs/operations/cut-over.md` (parity checklist per route + THROW paths + audit-username check, secrets seed via the `secret` console + rotation, NSSM side-by-side on `:9877`, port swap to `:9876`, Overview registry entry), `deploy/` (`publish.ps1`, `install-service.bat` / `uninstall-service.bat` for NSSM service `ZB.MOM.WW.BTDB.Manager`), `tools/smoke-new-queries.sql` (runs the six never-executed queries read-only against DEV with sqlcmd). Still open after cut-over: Overview lacks the mockup's per-machine "open > 24 h" rows; Ldap `Server`/`SearchBase`/`ServiceAccountDn` are blank in `appsettings.json` by design (per-environment).
+
+## 9. Local Docker rig — `docker/` (2026-09-04)
+
+Mirrors the HistorianGateway / Overview pattern: publish framework-dependent on the host (Gitea feed creds), copy into `aspnet:10.0`, never restore in Docker. Compose project `zb-btdb`, container `zb-btdb-manager`, image `zb-btdb-manager:local`, port `:9877` (same as the DEV side-by-side port, so `appsettings.json`'s Kestrel section is the single binding authority; the Dockerfile clears `ASPNETCORE_URLS`/`ASPNETCORE_HTTP_PORTS`). `ASPNETCORE_ENVIRONMENT=Production`; all environment-specific values are compose env vars, dev-only and committed under the family's rig exception.
+
+- **Database:** no MSSQL of its own — joins the external `scadabridge-net` and uses ScadaBridge's `scadabridge-mssql` (the instance that already carries `ScadaBridgeConfig` + `ZbSecretsHub`). `docker/deploy-db.sh`: (1) `docker/db/BT.Clr.csproj` compiles `Database/**/*.cs` to `Database.dll` (net472 via the auto-referenced reference assemblies — builds on macOS); (2) `docker/db/BT.Local.sqlproj` (SDK-style `Microsoft.Build.Sql/2.2.0`, `EnableDefaultSqlItems=false`, globs the four `Database/` folders, Sql130 model, suppresses the legacy build's own SQL71562/71558/71502) → `BT.Local.dacpac`; (3) `db/setup.sql` — `BT` + login `btdb_app` (db_owner); (4) `db/clr-objects.sql` — `clr enabled` on the instance, `sp_add_trusted_assembly` by SHA2-512 (strict security stays on), `CREATE ASSEMBLY [Database] … SAFE` from the bytes streamed as 2000-char `0x` chunks (a single 57 KB literal gets split by sqlcmd), then `StringJoin` aggregate, `StringCompare`/`CheckTriggered`/`WhenTriggered`/`WhenCleared`/`GetAggregateValues`, `GetTSV`/`GetMachineBatchTSV`/`GetCycleTSV`/`GetLinkedBatchTSV` with SSDT's type mapping; (5) `sqlpackage /Action:Publish` from a one-shot container (`db/Dockerfile.sqlpackage`, SDK image + the dotnet tool — the host's 10.0.5 runtime could not run the tool's 10.0.11 pin) with `DropObjectsNotInSource=false`, `BlockOnPossibleDataLoss=false`, `AllowIncompatiblePlatform=true`; (6) smoke. Result on the rig: 23 tables, 37 T-SQL procs + 1 TVF, 10 CLR objects, `Database` assembly `SAFE_ACCESS`. The script diffs the legacy `<Build Include>` list against the folders and warns on drift (found: `ElmaData_Nitric.sql` + `Elma_Bulk_Query_Nitric.sql` were on disk but not in `Database.sqlproj` — added). `ScadaBridge/infra/teardown.sh`/`reseed.sh` drop the volume and BT with it; re-run `deploy-db.sh`.
+- **Why not convert `Database.sqlproj` itself:** `Microsoft.Build.Sql` does not compile SQL CLR, and the Manager depends on `dbo.StringJoin` (batch search) and `dbo.StringCompare` (machine saves). The legacy SSDT project stays the artifact of record; the two side projects reference the same files, nothing is copied.
+- **Login:** enabled, `host.docker.internal:3893` + `FallbackServers` `:3894` (glauth-redundant), bind `cn=serviceaccount,dc=zb,dc=local`; `GroupToRole` `SCADA-Admins→Administrator`, `SCADA-Designers→Operator`, `SCADA-Viewers→Viewer`. Verified: `admin`/`designer` sign in, `gw-viewer` refused (fail closed). Literal `ConnectionStrings__BT` / LDAP password env vars override the `${secret:}` refs, so the `/data/btdb-secrets.db` store stays empty and no master key is set (confirmed to boot without one). The two config-import sources point at the local BT.
+- **Health / Overview:** `/healthz` 200, `/health/ready` 200 with the `sql` entry (503 when BT is down), `/health/active` 200, `/metrics`. Compose `healthcheck` = bash `/dev/tcp` GET of `/healthz` (liveness on purpose). Registered in `ZB.MOM.WW.Overview/src/…/appsettings.Docker.json` as application **BTDB Manager**, instance `manager`, `BaseUrl http://zb-btdb-manager:9877`, `HasActiveRole false`; the containerised Overview was republished and shows the tile Up with `sql` healthy (the dashboard already sits on `scadabridge-net`, so no extra network).
+
+
+## 10. State as of 2026-09-11 (supersedes §1, §4, §6.1, §6.3, §6.7, §8 where they differ)
+
+- **Everything is on local `master`** (`78195b8`): the Blazor conversion (§8) was merged, the
+  `feat/manager-blazor` worktree is gone, and the 2026-09-04 full review
+  (`docs/reviews/2026-09-04-full-review.md`, reviewed at `c624d41`) was remediated in 12 work-package
+  branches — `fix/save-batch-edit-null-compare`, `fix/db-lifecycle-sprocs`, `fix/clr-openquery-escaping`,
+  `fix/config-import-parents-and-nulls`, `test/db-integration`, `fix/manager-robustness`,
+  `fix/auth-hardening`, `ops/runbook-and-deploy`, `perf/indexes`, `db/schema-hygiene`,
+  `fix/concurrency-and-async`, `chore/nits` — all merged; 60 findings fixed, 3 partial by decision
+  (DB-15 `QV_*` impersonation, DB-16 alarm-CTE consolidation, DB-21 ServiceStack table drop), 0 open.
+  `feat/machine-batch-transfer` (DARS sprint 1, §1 item 1) is merged too, and so is
+  `feat/transfer-machine-batch` (merged `be976a6`, 2026-09-11): **`dbo.TransferMachineBatch`**, a tenth
+  Galaxy-facing writer that ends the source batch if still open, starts the destination batch by machine
+  `Code`, copies the `MachineBatchWO` rows still on the source (`MoveOutTimestamp IS NULL`, narrowed to
+  `@ContainerID`; blank = whole batch; source rows never modified) and records the `MachineBatchTransfer`
+  link — by composing `EndMachineBatch` / `StartMachineBatch` / `AddMachineBatchTransfer` in one owned
+  transaction. Outputs `@o_DestMachineBatchID`, `@o_MachineBatchTransferID`. Documented in
+  `docs/BatchTracker_Reference.md`; proven by `ValidateLifecycleSprocs.sql` §F (82 PASS on the rig).
+  `fix/transfer-lot-walk` (merged `5f0da01`, 2026-09-11) **reopened review DB-6**: the 2026-09-04 loop guard
+  in `AddMachineBatchTransfer` walked backward (toward the batches that fed the destination), which the
+  already-transferred check makes unreachable for any source, so `A→B` then `B→A` was still accepted; it
+  now walks forward along source→dest links with `HOLDLOCK`, and the destination row is held like the
+  source. The chain walks in `GetLinkedBatchTSV.cs` and `LinkedBatchAlarm_Wrapper.sql` picked one root
+  with `TOP 1` and dropped the other branch of a lot merge (`A→C, B→C`); both now resolve the whole lot
+  (forward to the final batch, then every batch that fed it) with one CTE pair pinned equal by
+  `SchemaHygieneTests`. Rig proofs: `ValidateLifecycleSprocs.sql` 87/0, `docker/db/validate-clr.sql`
+  27/0 over a merged lot with a `Runtime.dbo.Events` stand-in.
+  `feat/manager-transfers` (merged `695192d`, 2026-09-11) gave the **Manager** the transfer side it lacked: the
+  batch edit page edits the batch's outgoing `MachineBatchTransfer` row as a fourth MERGEd collection in
+  `SaveBatchEdit.sql` (guards `50015` self, `50016` destination missing, `50017` loop, `50018` more than one;
+  destination row and chain walk held like `AddMachineBatchTransfer`), audited into the new
+  `AuditLogMachineBatchTransfer` under the batch-edit header (new TVP `MachineBatchTransferType`); a read-only
+  lot panel (`GetBatchLot.sql`, the same `Down`/`Lot` CTE as the CLR, pinned equal by a static test); a lineage
+  column on Find Batch; a Transfers section on the batch-edit audit page and PDF. Rig proofs: 28 integration
+  tests (three new: round trip with audit, loop refused, merged lot), smoke 17/17, lifecycle 87/0, transfer
+  chains 4/0, validate-clr 27/0; gate 961 tests, 0 warnings. No branch is in flight.
+- **`master` is ~198 commits ahead of `origin/master` (`d147c10`) and unpushed.** Pushing is the user's
+  call — the remote is the corporate `ZimmerBiomet/BTDB`.
+- **Solution shape now:** `BTDB.slnx` = `src/ZB.MOM.WW.BTDB.Manager` (Blazor host) +
+  `src/ZB.MOM.WW.BTDB.Manager.Core` (Dapper data layer, embedded `Queries/*.sql`) + five test projects
+  (`Manager.Core.Tests` static query invariants, `Manager.Tests` bUnit/boot, `Database.Tests` static SQL
+  + sqlproj invariants, `Clr.Tests` compiles the CLR sources into net10, `Manager.Core.IntegrationTests`
+  opt-in on the rig behind `BTDB_TEST_CONNECTION`). `BTDB.sln` keeps `Database/Database.sqlproj`
+  (Windows MSBuild only). `.gitea/workflows/build.yml` runs `./build.sh` (needs `GITEA_NUGET_TOKEN`).
+- **Invariants a future session must not break** (all test-enforced): TVPs bind by ordinal so model
+  property order + lengths must equal `Database/Types/*.sql`; every write query is transaction-free and
+  audited (the C# wrapper owns the transaction); `dbo.X` only, never `BT.dbo.X`; `SYSUTCDATETIME()` in
+  Manager queries; every writer sproc follows the own-or-join `TRY/CATCH` template with locked lookups;
+  every `.sql` on disk must be in `Database.sqlproj`'s `<Build Include>` list; optimistic concurrency via
+  `ExpectedLastUpdate` → `THROW 50014`.
+- **Secrets/credentials (§6.2):** `${secret:btdb/connectionstrings/{bt,qa,prod}}` (the LDAP
+  service-account secret is gone since 2026-10-01 — see below; only `dotnet run` against local glauth uses it), store at `C:\ProgramData\ZB.MOM.WW\btdb-manager\secrets.db`,
+  master key only in `ZB_SECRETS_MASTER_KEY`. Rotation is cut-over step 4.2.
+- **Claude artifacts are gitignored + untracked in btdb** (2026-09-11): `CLAUDE.md`, `.claude/`,
+  `docs/plans/`, `docs/reviews/`, `*.tasks.json` — they exist on disk (btdb's `CLAUDE.md` is the repo
+  orientation and still loads) but never reach the corporate remote. `docs/operations/`,
+  `docs/BatchTracker_Reference.md`, `docs/mockups/` stay tracked.
+- **Open step:** phase-4 cut-over on the DEV box (`docs/operations/cut-over.md`; 4.1 parity walk-through
+  gates the rest) — **blocked by the 2026-09-11 remote-access hold** (no psbridge, no winrm; UNC copies and
+  direct SQL logins still allowed). §6 item 6 (which SQL host carries DEV `BT`) is still unrecorded.
+- **2026-10-01 — LDAP sign-in by user bind** (merged `a81a480`): Auth 0.4.0/0.4.1 added opt-in
+  `LdapOptions.BindMode=User`; BTDB ships it. Sign-in binds as `UserBindTemplate` (`<DOMAIN>\{0}` —
+  down-level maps exactly to the sAMAccountName the group search uses), the same connection reads the
+  user's own `memberOf`/`displayName`, and the username returned is the directory's spelling. No service
+  account, no LDAP secret; the preflight requires the template and refuses `AllowInsecure`/`Transport=None`
+  with user bind outside Development. Local dev + docker rig stay `BindMode=ServiceAccount` (GLAuth bind DNs
+  embed the user's OU). Rollout to the three Manager nodes (vd03 first) and retiring the stored secret are
+  the open steps (plan `docs/plans/2026-10-01-ldap-user-bind.md`, gitignored).
@@ -0,0 +1,5 @@
+# ScadaBridge — index history
+
+- **2026-10-13 existing fixes verified and issues closed.** Main `513fa1aa` already includes #35 fail-closed unavailable configured body/SQL redactors (`006202f3`) and #37 Flow Debugger structured/raw validated method inputs (`b5eb08f1`). Fresh full AuditLog project 407 passed/0 failed; FlowDebuggerPageTests 47 passed/0 failed; both terminal exit0. Remote ancestry confirmed; #35/#37 closed with evidence. No source changes/deployment/MES transactions. #36 runtime reset-echo overload is implemented, but QA/prod script port remains tracked; #30 per-site PSK deployment gate remains open.
+
+- **2026-10-13 #36 runtime regression proof refreshed.** Rebuilt HandshakeResetEchoTests passed all three cases with local fakes, terminal exit0; an earlier no-build attempt matched no tests and is excluded from evidence. Tracker now distinguishes implemented runtime b623fdb0 from remaining QA/prod database-owned script rollout (deployment register row119, scripts13/14/36 and15/16/25/26/37/38). No database writes, deployments or fresh live rig claims.
@@ -7,8 +7,10 @@ obj/
 [Aa]rtifacts/
 *.user

-# Claude Code tooling / agent worktrees (local-only)
-.claude/
+# Claude Code tooling / agent worktrees (local-only) — but .claude/skills/ IS tracked:
+# the sister-project-history skill holds the per-project narratives the root index points at.
+.claude/*
+!.claude/skills/

 # Local ops/infra notes — internal access details, never publish
 zimmer-vpn-tunnel.md
@@ -0,0 +1,37 @@
+# ZB.MOM.WW.Audit
+
+Shared audit event model + writer seam for the **ZB.MOM.WW SCADA family** (single package; DEEP-adopted by all apps as the canonical record). Build/test from this directory: `dotnet test`. Component spec + gaps: `../components/audit/`.
+
+## Index history (migrated from the scadaproj root CLAUDE.md)
+
+> Moved verbatim out of the scadaproj root `CLAUDE.md` on 2026-09-03 so it loads only when needed. **This file is now the authoritative place for this history — append new entries here and keep the root index row short.**
+
+### Narrative
+
+The audit component is fully populated: a normalized [`spec`](components/audit/spec/SPEC.md), an
+[`event-model`](components/audit/spec/EVENT-MODEL.md) reference, a
+[`shared-contract`](components/audit/shared-contract/ZB.MOM.WW.Audit.md), three
+[`current-state`](components/audit/current-state/) docs, and an adoption [`GAPS`](components/audit/GAPS.md)
+backlog. Common ground = canonical `AuditEvent` record + `AuditOutcome` enum + `IAuditWriter` /
+`IAuditRedactor` seams + helpers (`NullAuditRedactor`, `TruncatingAuditRedactor`, `NoOpAuditWriter`,
+`CompositeAuditWriter`, `RedactingAuditWriter`) + `AddZbAudit` DI registration; left per-project =
+transport/storage and domain vocabulary. Closes the loop on Auth — audit's `Actor` field = the Auth
+principal. `IAuditRedactor` is aligned with Telemetry's `ILogRedactor` seam convention.
+
+The shared library is **built and lives in this repo** at [`ZB.MOM.WW.Audit/`](ZB.MOM.WW.Audit/)
+(.NET 10; 1 package — `ZB.MOM.WW.Audit`; only non-BCL dependency `Microsoft.Extensions.DependencyInjection.Abstractions`;
+23 tests (verified 2026-07-18); `dotnet pack` → 1 nupkg @ 0.1.0). Repo: `https://gitea.dohertylan.com/dohertj2/zb-mom-ww-audit`.
+**Adopted across all three apps on 2026-06-02** (audit GAPS #1–#6) on each repo's `feat/adopt-zb-audit` branch
+(stacked on `feat/adopt-zb-auth`) — committed + reviewed, then **merged into the repo's local default (main/master)
+and PUSHED to origin (gitea) on 2026-06-03** (in sync). Depth =
+**DEEP adopt** (the canonical 9-field `AuditEvent` is the record everywhere; domain fields ride in `DetailsJson`).
+OtOpcUa: canonical record + `AuditWriterActor : IAuditWriter` + `Outcome` column/migration + `ClusterAudit` fix.
+MxGateway: new canonical SQLite `audit_event` store + `IAuditWriter` + `IApiKeyAuditStore`→canonical adapter.
+**ScadaBridge: a full audit-subsystem re-architecture** (the program's largest task) — canonical record everywhere via a
+deterministic codec; site SQLite split into `audit_event` + an `audit_forward_state` forwarding sidecar; central
+partitioned `dbo.AuditLog` collapsed to 10 canonical cols + persisted computed cols (`CollapseAuditLogToCanonical`
+migration, MSSQL-verified). Phase 3 wires `Actor` from the Auth principal at authenticated emit sites (per-app
+`IAuditActorAccessor`). Per-repo detail in [`components/audit/GAPS.md`](components/audit/GAPS.md) +
+`docs/plans/2026-06-02-auth-audit-normalization-phase2-deep.md` + `…-scadabridge-audit-rearch.md`.
+Build/test from `ZB.MOM.WW.Audit/`: `dotnet test`. Consumer matrix: all three apps consume the single
+`ZB.MOM.WW.Audit` package (OtOpcUa, MxAccessGateway, ScadaBridge — DEEP-adopted as the canonical record).
@@ -16,9 +16,9 @@ Canonical audit event model, best-effort writer seam, and redactor seam for the

 | Consumer | ZB.MOM.WW.Audit |
 |---|:---:|
-| **OtOpcUa** | yes (adoption deferred) |
-| **MxAccessGateway** | yes (adoption deferred) |
-| **ScadaBridge** | yes (adoption deferred — "align, don't replace") |
+| **OtOpcUa** | yes (adoption deferred; stale → #17) |
+| **MxAccessGateway** | yes (adoption deferred; stale → #17) |
+| **ScadaBridge** | yes (adoption deferred — "align, don't replace"; stale → #17) |

 Adoption is tracked in `components/audit/GAPS.md` in the outer `scadaproj` workspace. Each app brings its own transport (Akka broadcast / SQLite append / SQL ingest) and domain vocabulary (channels / kinds / event-types) — those stay per-project. The shared library provides the canonical record and the two seams that decouple "what to audit" from "how to store it".

@@ -1,4 +1,10 @@
 #!/usr/bin/env bash
 # pack.sh — produce the ZB.MOM.WW.Audit NuGet package into ./artifacts.
 set -euo pipefail
-dotnet pack -c Release -o ./artifacts
+# cd to the repo root and name the solution + an absolute output path: `dotnet pack` with
+# neither a project argument nor a fixed working directory packs whatever solution happens
+# to sit in the CALLER's directory (the bug behind the LocalDb feed incident, 0038d46).
+REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+cd "$REPO_ROOT"
+
+dotnet pack ZB.MOM.WW.Audit.slnx -c Release -o "$REPO_ROOT/artifacts"
@@ -16,9 +16,20 @@ set -euo pipefail
 : "${GITEA_NUGET_SOURCE:?set GITEA_NUGET_SOURCE to your Gitea NuGet feed URL}"
 : "${GITEA_NUGET_KEY:?set GITEA_NUGET_KEY to your Gitea access token}"

-dotnet pack -c Release -o ./artifacts
+# Named solution, named output, and a cd to the repo root, all for one reason: `dotnet pack`
+# with neither a project argument nor a fixed working directory packs whatever solution
+# happens to sit in the CALLER's directory — and then pushes it to a live feed under this
+# script's name. Invoked by absolute path from another repo's checkout, the relative form
+# packed and published 25 unrelated packages (LocalDb's copy of this script, fixed in
+# 0038d46; recovered with `dotnet nuget delete`, but only because the token happened to
+# carry delete rights). A publish script must be unambiguous about WHAT it publishes no
+# matter where it is run.
+REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+cd "$REPO_ROOT"

-dotnet nuget push "./artifacts/*.nupkg" \
+dotnet pack ZB.MOM.WW.Audit.slnx -c Release -o "$REPO_ROOT/artifacts"
+
+dotnet nuget push "$REPO_ROOT/artifacts/*.nupkg" \
  --source "$GITEA_NUGET_SOURCE" \
  --api-key "$GITEA_NUGET_KEY" \
  --skip-duplicate
@@ -0,0 +1,43 @@
+# ZB.MOM.WW.Auth
+
+Shared login / identity / API-key library for the **ZB.MOM.WW SCADA family** (4 packages + `ApiKeys.SqlServer`; consumed by OtOpcUa, ScadaBridge, MxAccessGateway, HistorianGateway). Build/test from this directory: `dotnet test` (the `ApiKeys.SqlServer` live suite is gated on `AUTH_SQLSERVER_CONNSTR`). Component spec + gaps: `../components/auth/`.
+
+**Backlog / open work:** lives in Gitea issues on `dohertj2/scadaproj` (`tea issues list --repo dohertj2/scadaproj`) — currently #5 (roll apps to `0.4.1`), #6 (VD03 NAM AD cutover), #18 (Secrets migrator jitter, ported from here). The history below is a record, not a status tracker; file new follow-ups as issues.
+
+## Index history (migrated from the scadaproj root CLAUDE.md)
+
+> Moved verbatim out of the scadaproj root `CLAUDE.md` on 2026-09-03 so it loads only when needed. **This file is now the authoritative place for this history — append new entries here and keep the root index row short.**
+
+### Component-table row as it read in the index
+
+| Component | Status | Goal | Design | Implementation |
+|---|---|---|---|---|
+| Auth (login / identity / authz) | Adopted (**feed carries `0.3.0` as of 2026-09-02 — five packages, the new one being `ZB.MOM.WW.Auth.ApiKeys.SqlServer`; ALL FOUR apps pin `0.3.0` as of 2026-09-03**: OtOpcUa `master @ b8b4c18f`, ScadaBridge `main @ e45c4af9` (the `inbound-api-key-shared-store` work, merged), MxAccessGateway `main @ 4f931e4`, HistorianGateway `main @ d8d890c` — each built + tested with no new failures versus its pre-bump baseline, all PUSHED to origin. The preceding `0.2.1` pins landed 2026-08-13 (OtOpcUa `08e33fc7`, ScadaBridge `d28824d9`, MxAccessGateway `1cb14d2`, HistorianGateway `e42ff31`). **`0.2.0` (2026-08-10) adds LDAP backup-DC sticky failover**, **`0.2.1` (2026-08-12) fixes a defect that broke AD logins outright** — an AD continuation referral escaped `NovellLdapConnection.Search`, so any `SearchBase` at a domain root failed every login as `ServiceAccountBindFailed` against a perfectly healthy directory. Live-proven against `DC=nam,DC=zimmer,DC=com`; the app-level exposure is closed by the pin bumps. **`FallbackServers` is live-gated** via the local redundant GLAuth pair ([`infra/glauth-redundant/`](infra/glauth-redundant/), primary :3893 + backup :3894 on the dev Mac): HistorianGateway dashboard drill PASS 2026-08-13 (failover, stickiness, walk-back). | Shared `ZB.MOM.WW.Auth` lib | [`components/auth/`](components/auth/) | [`ZB.MOM.WW.Auth/`](ZB.MOM.WW.Auth/) |
+
+### Narrative
+
+The auth component is fully populated: a normalized [`spec`](components/auth/spec/SPEC.md), a
+proposed [`shared-contract`](components/auth/shared-contract/ZB.MOM.WW.Auth.md), three
+[`current-state`](components/auth/current-state/) docs, and an adoption [`GAPS`](components/auth/GAPS.md)
+backlog. Common ground = LDAP/GLAuth identity + peppered-HMAC API keys; left per-project =
+the authz vocabularies (OPC-UA permissions / gRPC scopes / roles + site-scoping).
+
+The shared library is **built and lives in this repo** at [`ZB.MOM.WW.Auth/`](ZB.MOM.WW.Auth/)
+(a plain directory of this repo — no nested `.git`, verified 2026-08-07; .NET 10; 4 packages — `Abstractions`, `Ldap`, `ApiKeys`, `AspNetCore`;
+250 tests passing + 2 skipped (verified 2026-08-10 at `0.2.0`; was 215+1 at `0.1.5`); feed carries `0.1.0`–`0.2.0`). The implementation plan is at
+[`docs/plans/2026-06-01-zb-mom-ww-auth-shared-library.md`](docs/plans/2026-06-01-zb-mom-ww-auth-shared-library.md).
+**Adopted across all three apps on 2026-06-02** (auth GAPS #1–#8) on each repo's `feat/adopt-zb-auth` branch —
+committed + reviewed, then **fast-forward-merged into the repo's local default (main/master) and PUSHED to origin
+(gitea) on 2026-06-03** (in sync; the `feat/*` branches kept locally as history). Cutover: shared `Auth.Ldap`,
+`Auth.ApiKeys` (ScadaBridge inbound fully re-architected to the keyId/Bearer model), `IGroupRoleMapper<TRole>` seam,
+`Transport`-enum config, canonical `ZbClaimTypes`/`ZbCookieDefaults`, unified dev base DN `dc=zb,dc=local`, and the
+canonical-six role vocabulary (with ScadaBridge's accepted auditor/admin SoD collapse). Consumer pins: **all four apps pin `0.3.0` as of 2026-09-03** (commits listed in the table row above, all pushed to origin); all four pinned `0.2.1` from 2026-08-13 before that. Prior history — all four pinned `0.1.5` (verified 2026-08-07): MxGateway via `e107019` 2026-07-18 (earlier bumped `0.1.2`→`0.1.4` 2026-07-09 for the G-2 API-key-expiry verifier enforcement — commit `197731a`; also required implementing the 0.1.3 `IApiKeyAdminStore.SetScopes/SetEnabled` members on a test fake), ScadaBridge `50d79ed1` + HistorianGateway `df211ee` 2026-07-18, OtOpcUa `1adaa2fc` 2026-07-20 (PR #484, family alignment from `0.1.1`). Per-repo detail in [`components/auth/GAPS.md`](components/auth/GAPS.md) +
+`docs/plans/2026-06-02-auth-audit-normalization*.md`.
+Build/test from `ZB.MOM.WW.Auth/`: `dotnet test` (the `ApiKeys.SqlServer` live suite is gated on `AUTH_SQLSERVER_CONNSTR`, skipped otherwise). Consumer matrix: OtOpcUa → Abstractions+Ldap+AspNetCore;
+MxAccessGateway & HistorianGateway → the original four; ScadaBridge → those four **plus `ApiKeys.SqlServer`** (ApiKeys not used by OtOpcUa).
+**`0.1.4` (2026-07-09, archreview G-2):** optional `ApiKeyRecord.ExpiresUtc` (nullable = never-expires), `ApiKeyFailure.KeyExpired`, `ApiKeyVerifier` expiry enforcement (inclusive, injected clock), a `CreateKeyAsync(expiresUtc)` overload, and a SQLite **schema v3** additive `expires_utc` column (idempotent `ALTER`; donor v2 DBs upgrade in place, no key invalidated). **All four Auth pkgs published to the Gitea feed at `0.1.4`** (2026-07-09). **HistorianGateway adopted the expiry surface at `0.1.4`** — `apikey create --expires` + dashboard Expires column (Task-10 done). **`0.1.5` (2026-07-18)** is `0.1.4` plus the `SQLitePCLRaw.lib.e_sqlite3 2.1.12` transitive security pin (GHSA-2m69-gcr7-jv3q), no API change; all four apps now pin it (see consumer pins above). **`0.2.0` (2026-08-10) — LDAP backup-DC sticky failover** (plan `docs/plans/2026-08-10-ldap-backup-dc-failover.md`, design `…-design.md`; **4 pkgs published to the feed, restore-verified**; family `Directory.Build.props` now `0.2.0`): one additive `LdapOptions.FallbackServers` list (`"host"` or `"host:port"`, empty default = failover off, single-endpoint behaviour byte-unchanged) makes `LdapAuthService` walk an ordered endpoint list (primary + fallbacks) on **system-side failures only** — connect/TLS, service-account bind, or search. A user-side answer from a healthy directory (`Success`/`UserNotFound`/`AmbiguousUser`/`BadCredentials`/`GroupLookupFailed`) is **final and never retried elsewhere** (re-asking a second DC about a rejected password would multiply one bad-password event into several against the account lockout policy). Endpoint preference is **sticky** — the last endpoint that gave *any* authoritative answer (success **or** a user-side failure) is tried first next time — so a primary outage costs its connect timeout once, not per login; an exhausted walk leaves the preference untouched. Fail-closed contract preserved per-endpoint; boot validation rejects a malformed `FallbackServers` entry naming the field; optional `ILogger<LdapAuthService>` emits a Warning per failed hop (endpoint + stage, no secrets) and an Information on sticky change; wired via `AddZbLdapAuth` (`GetService`, so logging is optional). Motivation: the ZB production directory is NAM AD with two DCs. Unit suite covers the walk/stickiness (14 tests) + env-gated live GLAuth failover test (`ZB_LDAP_IT=1`). **No `Abstractions` contract change beyond the new option; no `LdapAuthFailure` enum change.** App adoption (pin bump to `0.2.0` + authoring `FallbackServers`) is a follow-up.
+**`0.2.1` (2026-08-12) — AD continuation referrals no longer fail the search** (**4 pkgs published to the feed, restore-verified**; the assembly restored *from the feed* carries the fix and `0.2.0` does not — checked as a positive control, not by trusting the push). Active Directory answers **any search based at a domain root** with continuation references (to `CN=Configuration`, `DC=ForestDnsZones`, …). With referral following disabled — the default, and what we want, since chasing one would re-bind somewhere we never authenticated — Novell surfaces each by throwing `LdapReferralException` from `Next()`. That escaped `NovellLdapConnection.Search`, and `LdapAuthService` maps a failed search to `ServiceAccountBindFailed`, so **every login against `SearchBase = DC=nam,DC=zimmer,DC=com` failed system-side on both DCs while the directory, the credentials and the service account were all healthy** — ~546 ms, identical on each endpoint, so failover could not help either. Found by the new `tools/LdapProbe` (below) and live-proven fixed against the ZB NAM domain root. The search loop now skips referrals (`ReadEntries(ILdapSearchResults)`, extracted so it is unit-testable through the Novell seam without a live directory) and continues; a 1000-referral cap guards a client that throws without advancing, since that would spin forever inside a fail-closed auth path. **Non-referral faults still propagate** — swallowing those would turn "the directory is broken" into "no such user", which is the wrong answer to give about someone's credentials. 7 new tests (`NovellSearchReferralTests`), **6 verified red against the pre-fix code**, the seventh being the non-referral control that must stay green; suite 257 pass / 2 skip / 0 fail. **No `Abstractions` change; no config change.** **Consumer impact: an app on `0.1.5` cannot authenticate against an AD domain root at all**, so the pin bump is a functional fix, not housekeeping. **Closed 2026-08-13: all four apps adopted `0.2.1`** (see the table row for commits), and `FallbackServers` was live-gated the same day against the redundant local GLAuth pair (`infra/glauth-redundant/`).
+
+**`0.3.0` (2026-09-02) — `ZB.MOM.WW.Auth.ApiKeys.SqlServer`, a SQL Server provider for the API-key store** (design `ScadaBridge/docs/plans/2026-09-02-inbound-api-key-shared-store-design.md`; **5 pkgs published to the feed**, `ApiKeys.SqlServer` search-verified from the ScadaBridge checkout; family `Directory.Build.props` now `0.3.0`). Motivation: `Auth.ApiKeys`' SQLite store is a per-node file, so on a two-node active/standby host (ScadaBridge central) a key minted on one node exists only there and a failover 401s every inbound caller (ScadaBridge deferred-work row 108). The new package implements `IApiKeyStore`/`IApiKeyAdminStore`/`IApiKeyAuditStore` over `Microsoft.Data.SqlClient` in a `zbauth` schema (v1: `schema_version`, `api_keys`, `api_key_audit`; all timestamps `DATETIMEOFFSET(7)`, tick-exact) with a concurrent-boot-safe migrator (Serializable tx, retry on 1205/2714/1913/2627 with **jittered** backoff — the Secrets replicator's flat `100*attempt` backoff exhausted five retries about once per sixty 8-way boots in the port's soak test, so the same latent defect is an open follow-up there → #18) and a `SqliteApiKeyStoreImporter` that copies an existing SQLite file into the schema insert-where-absent (idempotent, so both nodes of a pair can import their own file and the result is the union). Wiring: `services.AddZbApiKeyAuthSqlServer(config, apiKeySection, sqlServerSection)` registers the SQL Server stores + migrator **before** delegating to `AddZbApiKeyAuth`, whose `TryAdd*` registrations then yield (same displacement mechanism as Secrets 0.5.0). One `Abstractions` addition, additive: `IApiKeyStoreMigrator`, which `SqliteAuthStoreMigrator` now implements and `ApiKeyAdminCommands` accepts via a new ctor (old `SqliteAuthStoreMigrator` ctor retained — the three `0.2.1` consumers compile unchanged). Store semantics are pinned to the SQLite contract tests re-run live (mark-used skips revoked keys, delete only removes revoked keys, rotate clears last-used/revoked, list newest-first). 92 tests in the new project, 351 across the solution, 0 failed. A DBA-run copy of the DDL ships at `docs/sql/zbauth-schema.sql`.
+
+**`tools/LdapProbe`** (`ldapprobe`, not packable, outside the Auth solution) is the diagnostic that found the above: it drives the **real** `LdapAuthService` so its verdict is what the four apps produce, then separately queries the directory for what the library cannot report — full group DNs, **nested** groups via `LDAP_MATCHING_RULE_IN_CHAIN`, the **primary group** (which AD omits from `memberOf`, so it can never be a role group), group type (security vs distribution), decoded `userAccountControl`, referral counts, and the TLS chain with key sizes. `--list-groups '<pat>'` enumerates every group in the directory matching a pattern, which a membership dump structurally cannot do. Findings recorded in [`ldap_setup.md`](ldap_setup.md), including the **`UG_WW_*` Wonderware group convention (39 groups)** that partly overturns that doc's earlier "no group→role convention exists" claim — it was true of Ignition, false of the domain.
@@ -5,7 +5,12 @@
    <Nullable>enable</Nullable>
    <ImplicitUsings>enable</ImplicitUsings>
    <LangVersion>latest</LangVersion>
-    <Version>0.1.5</Version>
+    <!-- 0.4.0: opt-in LdapOptions.BindMode = User — authenticate by binding as the signing-in user
+         (UserBindTemplate) and read their own groups on the same connection; no service account.
+         ServiceAccount stays the default, so existing consumers need no change. -->
+    <!-- 0.4.1: Username canonicalised from the entry's UserNameAttribute; empty passwords refused
+         in both bind modes (ServiceAccount-mode behaviour change); template docs prefer EXAMPLE\{0}. -->
+    <Version>0.4.1</Version>
    <ManagePackageVersionsCentrally>true</ManagePackageVersionsCentrally>
  </PropertyGroup>

@@ -15,6 +15,14 @@
    <!-- Transitive security pin: Microsoft.Data.Sqlite 10.0.7 pulls SQLitePCLRaw.lib.e_sqlite3 2.1.11,
         which carries advisory GHSA-2m69-gcr7-jv3q (NU1903). Pin the patched 2.1.12 native lib. -->
    <PackageVersion Include="SQLitePCLRaw.lib.e_sqlite3" Version="2.1.12" />
+    <!-- Shared SQL-Server API-key store (ZB.MOM.WW.Auth.ApiKeys.SqlServer). 6.0.2 matches the
+         family pin used by ZB.MOM.WW.Secrets' SQL-Server store, so a consumer taking both gets one
+         SqlClient rather than a downgrade warning. -->
+    <PackageVersion Include="Microsoft.Data.SqlClient" Version="6.0.2" />
+    <!-- 9.0.4, not 9.0.0: Microsoft.Data.SqlClient 6.0.2 pulls System.Configuration.ConfigurationManager
+         9.0.4, which requires ProtectedData >= 9.0.4. With transitive pinning on, the lower central
+         version is a hard NU1109 downgrade error rather than a silent resolve. -->
+    <PackageVersion Include="System.Security.Cryptography.ProtectedData" Version="9.0.4" />

    <!-- Extensions -->
    <PackageVersion Include="Microsoft.Extensions.Options" Version="10.0.7" />
@@ -24,6 +32,7 @@
    <PackageVersion Include="Microsoft.Extensions.Configuration.Abstractions" Version="10.0.7" />
    <PackageVersion Include="Microsoft.Extensions.Configuration" Version="10.0.7" />
    <PackageVersion Include="Microsoft.Extensions.Hosting.Abstractions" Version="10.0.7" />
+    <PackageVersion Include="Microsoft.Extensions.Logging.Abstractions" Version="10.0.7" />

    <!-- ASP.NET Core Authentication / Authorization -->
    <PackageVersion Include="Microsoft.AspNetCore.Authentication.JwtBearer" Version="10.0.7" />
@@ -11,17 +11,18 @@ Authentication and authorisation libraries for the **ZB.MOM.WW SCADA family** (O
 | `ZB.MOM.WW.Auth.Abstractions` | Auth contracts, canonical role constants, and shared types (`LdapOptions`, `LdapAuthResult`, `ILdapAuthService`, `IApiKeyStore`). No runtime dependencies beyond the BCL. | — |
 | `ZB.MOM.WW.Auth.Ldap` | LDAP authentication service: bind-then-search-then-bind against GLAuth or Active Directory; RFC 4514-aware group extraction; fail-closed. | `Abstractions`, `Novell.Directory.Ldap.NETStandard` |
 | `ZB.MOM.WW.Auth.ApiKeys` | SQLite-backed API-key store with **pepper-keyed HMAC-SHA256** secret hashing, rotation, and audit log. DI wiring is `AddZbApiKeyAuth`; an opt-in `MigrationHostedService` runs schema migrations on startup. | `Abstractions`, `Microsoft.Data.Sqlite` |
+| `ZB.MOM.WW.Auth.ApiKeys.SqlServer` | SQL Server-backed API-key store sharing **one row set** across clustered hosts (e.g. a central pair), instead of each node keeping its own SQLite file. DI wiring is `AddZbApiKeyAuthSqlServer(config, apiKeySection, sqlServerSection)`, called **instead of** `AddZbApiKeyAuth` (it calls that internally). Also ships `SqliteApiKeyStoreImporter` — a one-shot per-node SQLite → shared store copy a host runs once at boot. | `Abstractions`, `ApiKeys`, `Microsoft.Data.SqlClient` |
 | `ZB.MOM.WW.Auth.AspNetCore` | ASP.NET Core wiring for the **LDAP** provider only: `AddZbLdapAuth` (binds + start-time-validates `LdapOptions`, registers `ILdapAuthService`), plus `ZbCookieDefaults.Apply` (hardened cookie helper the consumer calls itself) and `ZbClaimTypes` constants. It does **not** wire API keys or cookie middleware — API-key DI is `AddZbApiKeyAuth` in the `ApiKeys` package. | `Abstractions`, `Ldap`, `Microsoft.AspNetCore.App` |

 ---

 ## Consumer Matrix

-| Consumer | Abstractions | Ldap | ApiKeys | AspNetCore |
-|---|:---:|:---:|:---:|:---:|
-| **OtOpcUa** | yes | yes | — | yes |
-| **MxAccessGateway** | yes | yes | yes | yes |
-| **ScadaBridge** | yes | yes | yes | yes |
+| Consumer | Abstractions | Ldap | ApiKeys | ApiKeys.SqlServer | AspNetCore |
+|---|:---:|:---:|:---:|:---:|:---:|
+| **OtOpcUa** | yes | yes | — | — | yes |
+| **MxAccessGateway** | yes | yes | yes | — | yes |
+| **ScadaBridge** | yes | yes | yes | yes (central pair, 2026-09-02) | yes |

 `ApiKeys` is NOT used by OtOpcUa (that app authenticates human operators via LDAP + cookies only; machine-to-machine access is out of scope).

@@ -29,7 +30,41 @@ Authentication and authorisation libraries for the **ZB.MOM.WW SCADA family** (O

 ## Versioning

-All four packages are versioned **lockstep** from `Directory.Build.props`. The current release is **0.1.4**. A single version bump in `Directory.Build.props` bumps all four packages simultaneously — consumers should reference the same version for all ZB.MOM.WW.Auth packages.
+All five packages are versioned **lockstep** from `Directory.Build.props`. The current release is **0.4.1**. A single version bump in `Directory.Build.props` bumps all five packages simultaneously — consumers should reference the same version for all ZB.MOM.WW.Auth packages.
+
+### 0.4.1 — LDAP review fixes
+
+- **Canonical username (both modes):** the search now also requests `UserNameAttribute`, and `LdapAuthResult.Username` is the entry's first value of it (the directory's spelling), so `JDoe` and `jdoe` sign in as one identity in consumers' audit columns. Falls back to the normalised typed name when the entry has no such value.
+- **Behaviour change for `ServiceAccount`-mode consumers:** an empty (or null) password is now refused by the library as `BadCredentials` before any network call, in both bind modes. Previously only `User` mode refused it; in `ServiceAccount` mode the re-bind as the user's DN with an empty password is an unauthenticated bind that AD reports as success, so a consumer without its own edge check admitted anyone.
+- **Template guidance:** prefer the down-level `UserBindTemplate` `EXAMPLE\{0}` — it maps exactly to the `sAMAccountName` the group search uses. A UPN template (`{0}@example.com`) is only correct where every account's UPN prefix equals its `sAMAccountName`; AD resolves UPN binds against the explicit `userPrincipalName`, so a mismatch can fail sign-in or bind a different account than the one whose groups are read.
+- No `Abstractions` contract change.
+
+### 0.4.0 — LDAP sign-in by user bind (opt-in)
+
+- **`LdapOptions.BindMode`** (`ServiceAccount` default | `User`), **`UserBindTemplate`** (`{0}` = the login name, e.g. `EXAMPLE\{0}` — preferred, see 0.4.1 — or `{0}@example.com`) and **`AcceptDomainQualifiedNames`** (default `true`).
+- In `User` mode `LdapAuthService` binds as the user; success authenticates them and the same connection reads their entry (`UserNameAttribute`, `DisplayNameAttribute`, `GroupAttribute`). No `ServiceAccountDn` / `ServiceAccountPassword` is needed or validated.
+- Failure mapping: empty password or an unusable name → `BadCredentials` with no network call (AD treats an empty-password simple bind as an unauthenticated success). Bind result 49 → `BadCredentials`, final — never retried on a fallback DC, so one wrong password is one lockout strike. Any other bind error, a connect failure or a search failure → the system-side bucket (`ServiceAccountBindFailed`) and the next endpoint. Zero/several entries or no groups after a good bind → `GroupLookupFailed`.
+- `DOMAIN\name` / `name@suffix` are reduced to `name` (or refused when `AcceptDomainQualifiedNames` is `false`), so the result's `Username` has one spelling (0.4.1 also canonicalises its case from the entry).
+- No `Abstractions` contract change beyond the new enum and option properties; `ServiceAccount` mode is unchanged.
+- `UserBindTemplate` output is used verbatim as the bind name, so use a down-level (`EXAMPLE\{0}`, preferred) or UPN (`{0}@example.com`, only where every UPN prefix equals the `sAMAccountName`) template; a DN template is not escaped.
+
+### 0.3.0 — SQL Server API-key store (ScadaBridge central pair)
+
+- **`IApiKeyStoreMigrator`** added to `Abstractions`: creates or upgrades an API-key store's schema; idempotent and safe to run concurrently from several nodes. `SqliteAuthStoreMigrator` implements it unchanged. `AddZbApiKeyAuth` registers the SQLite implementation via `TryAdd`, so a provider that registers its own migrator first (see below) wins — that ordering is the whole mechanism by which a different backend is swapped in.
+- **New package `ZB.MOM.WW.Auth.ApiKeys.SqlServer`** + `AddZbApiKeyAuthSqlServer(config, apiKeySection, sqlServerSection)`, called instead of `AddZbApiKeyAuth` (it calls it internally after registering the SQL Server stores and migrator first).
+- **`zbauth` schema v1** (three tables: `schema_version`, `api_keys`, `api_key_audit`; `DATETIMEOFFSET(7)` timestamps). `SqlServerApiKeyStoreMigrator` is concurrent-boot safe — a central pair's two nodes both migrate at every boot — with retry on SQL error numbers 1205/2714/1913/2627 and jittered backoff (measured: the flat backoff it replaced exhausted its retries roughly 1 in 60 eight-way boots). DBA-facing DDL: `docs/sql/zbauth-schema.sql`.
+- **`SqliteApiKeyStoreImporter`** — idempotent insert-where-not-exists copy of an existing node's per-node SQLite keys and audit rows into the shared store; a row already present in SQL Server is never modified, so running it from both nodes of a pair converges on the union of their files.
+- **`ApiKeyAdminCommands`** gained a ctor taking `IApiKeyStoreMigrator`; the existing ctor taking the concrete `SqliteAuthStoreMigrator` is retained (MxAccessGateway, HistorianGateway and ScadaBridge all construct it that way).
+- SQLite remains the default store; **existing consumers need no code change to move to 0.3.0**.
+- Live tests for the SQL Server package are gated on `AUTH_SQLSERVER_CONNSTR` and skip (not fail) when it is unset.
+
+### 0.2.1 — LDAP AD continuation-referral fix
+
+- `NovellLdapConnection.Search` now skips Active Directory continuation referrals (`LdapReferralException` from `Next()`) instead of letting one fail the whole search — a `SearchBase` at a domain root made every login fail as `ServiceAccountBindFailed` against a healthy directory. Non-referral faults still propagate untouched.
+
+### 0.2.0 — version bump only
+
+- No functional change; bumped the lockstep family version ahead of the 0.2.1 LDAP fix.

 ### 0.1.4 — API-key expiry (archreview G-2)

@@ -2,11 +2,13 @@
  <Folder Name="/src/">
    <Project Path="src/ZB.MOM.WW.Auth.Abstractions/ZB.MOM.WW.Auth.Abstractions.csproj" />
    <Project Path="src/ZB.MOM.WW.Auth.ApiKeys/ZB.MOM.WW.Auth.ApiKeys.csproj" />
+    <Project Path="src/ZB.MOM.WW.Auth.ApiKeys.SqlServer/ZB.MOM.WW.Auth.ApiKeys.SqlServer.csproj" />
    <Project Path="src/ZB.MOM.WW.Auth.AspNetCore/ZB.MOM.WW.Auth.AspNetCore.csproj" />
    <Project Path="src/ZB.MOM.WW.Auth.Ldap/ZB.MOM.WW.Auth.Ldap.csproj" />
  </Folder>
  <Folder Name="/tests/">
    <Project Path="tests/ZB.MOM.WW.Auth.ApiKeys.Tests/ZB.MOM.WW.Auth.ApiKeys.Tests.csproj" />
+    <Project Path="tests/ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests/ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests.csproj" />
    <Project Path="tests/ZB.MOM.WW.Auth.AspNetCore.Tests/ZB.MOM.WW.Auth.AspNetCore.Tests.csproj" />
    <Project Path="tests/ZB.MOM.WW.Auth.Ldap.Tests/ZB.MOM.WW.Auth.Ldap.Tests.csproj" />
  </Folder>
@@ -6,4 +6,10 @@

 set -euo pipefail

-dotnet pack -c Release -o ./artifacts
+# cd to the repo root and name the solution + an absolute output path: `dotnet pack` with
+# neither a project argument nor a fixed working directory packs whatever solution happens
+# to sit in the CALLER's directory (the bug behind the LocalDb feed incident, 0038d46).
+REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+cd "$REPO_ROOT"
+
+dotnet pack ZB.MOM.WW.Auth.slnx -c Release -o "$REPO_ROOT/artifacts"
@@ -16,9 +16,20 @@ set -euo pipefail
 : "${GITEA_NUGET_SOURCE:?set GITEA_NUGET_SOURCE to your Gitea NuGet feed URL}"
 : "${GITEA_NUGET_KEY:?set GITEA_NUGET_KEY to your Gitea access token}"

-dotnet pack -c Release -o ./artifacts
+# Named solution, named output, and a cd to the repo root, all for one reason: `dotnet pack`
+# with neither a project argument nor a fixed working directory packs whatever solution
+# happens to sit in the CALLER's directory — and then pushes it to a live feed under this
+# script's name. Invoked by absolute path from another repo's checkout, the relative form
+# packed and published 25 unrelated packages (LocalDb's copy of this script, fixed in
+# 0038d46; recovered with `dotnet nuget delete`, but only because the token happened to
+# carry delete rights). A publish script must be unambiguous about WHAT it publishes no
+# matter where it is run.
+REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+cd "$REPO_ROOT"

-dotnet nuget push "./artifacts/*.nupkg" \
+dotnet pack ZB.MOM.WW.Auth.slnx -c Release -o "$REPO_ROOT/artifacts"
+
+dotnet nuget push "$REPO_ROOT/artifacts/*.nupkg" \
  --source "$GITEA_NUGET_SOURCE" \
  --api-key "$GITEA_NUGET_KEY" \
  --skip-duplicate
@@ -0,0 +1,69 @@
+-- ZB.MOM.WW.Auth.ApiKeys.SqlServer — shared API-key schema, version 1.
+--
+-- For DBAs whose application login lacks DDL rights. The host normally provisions this itself at
+-- startup (SqlServerApiKeyStoreMigrator, which runs the identical statements and is safe to run
+-- concurrently from every node); where the app login may only read and write rows, run this script
+-- once against the target database instead. The migrator then finds the version row already at the
+-- version it supports and does nothing further beyond re-stamping applied_utc.
+--
+-- The schema name is hard-coded to zbauth here — the default of SqlServerApiKeyOptions.Schema. If a
+-- deployment overrides that option, substitute the name in every bracketed identifier below.
+--
+-- Kept byte-for-byte consistent with SqlServerAuthSchema.CreateSchemaDdl apart from that
+-- substitution and the literal version row at the end; change both together.
+
+IF NOT EXISTS (SELECT 1 FROM sys.schemas WHERE name = N'zbauth')
+    EXEC(N'CREATE SCHEMA [zbauth]');
+
+IF OBJECT_ID(N'[zbauth].[schema_version]', N'U') IS NULL
+    CREATE TABLE [zbauth].[schema_version] (
+        id          INT               NOT NULL CONSTRAINT PK_zbauth_schema_version PRIMARY KEY CHECK (id = 1),
+        version     INT               NOT NULL,
+        applied_utc DATETIMEOFFSET(7) NOT NULL
+    );
+
+IF OBJECT_ID(N'[zbauth].[api_keys]', N'U') IS NULL
+    CREATE TABLE [zbauth].[api_keys] (
+        key_id        NVARCHAR(64)      NOT NULL CONSTRAINT PK_zbauth_api_keys PRIMARY KEY,
+        key_prefix    NVARCHAR(16)      NOT NULL,
+        secret_hash   VARBINARY(64)     NOT NULL,
+        display_name  NVARCHAR(256)     NOT NULL,
+        scopes        NVARCHAR(MAX)     NOT NULL,
+        constraints   NVARCHAR(MAX)     NULL,
+        created_utc   DATETIMEOFFSET(7) NOT NULL,
+        last_used_utc DATETIMEOFFSET(7) NULL,
+        revoked_utc   DATETIMEOFFSET(7) NULL,
+        expires_utc   DATETIMEOFFSET(7) NULL
+    );
+
+IF OBJECT_ID(N'[zbauth].[api_key_audit]', N'U') IS NULL
+    CREATE TABLE [zbauth].[api_key_audit] (
+        audit_id       BIGINT IDENTITY(1,1) NOT NULL CONSTRAINT PK_zbauth_api_key_audit PRIMARY KEY,
+        key_id         NVARCHAR(64)      NULL,
+        event_type     NVARCHAR(64)      NOT NULL,
+        remote_address NVARCHAR(64)      NULL,
+        created_utc    DATETIMEOFFSET(7) NOT NULL,
+        details        NVARCHAR(MAX)     NULL
+    );
+
+IF NOT EXISTS (
+    SELECT 1 FROM sys.indexes
+    WHERE name = N'ix_api_keys_revoked_utc'
+      AND object_id = OBJECT_ID(N'[zbauth].[api_keys]'))
+    CREATE INDEX ix_api_keys_revoked_utc
+        ON [zbauth].[api_keys] (revoked_utc);
+
+IF NOT EXISTS (
+    SELECT 1 FROM sys.indexes
+    WHERE name = N'ix_api_key_audit_key_id_created_utc'
+      AND object_id = OBJECT_ID(N'[zbauth].[api_key_audit]'))
+    CREATE INDEX ix_api_key_audit_key_id_created_utc
+        ON [zbauth].[api_key_audit] (key_id, created_utc);
+
+MERGE [zbauth].[schema_version] AS target
+USING (SELECT 1 AS id, 1 AS version, SYSDATETIMEOFFSET() AS applied_utc) AS source
+    ON target.id = source.id
+WHEN MATCHED THEN
+    UPDATE SET version = source.version, applied_utc = source.applied_utc
+WHEN NOT MATCHED THEN
+    INSERT (id, version, applied_utc) VALUES (1, source.version, source.applied_utc);
@@ -78,3 +78,16 @@ public interface IApiKeyAuditStore
    Task AppendAsync(ApiKeyAuditEntry entry, CancellationToken ct);
    Task<IReadOnlyList<ApiKeyAuditEntry>> ListRecentAsync(int limit, CancellationToken ct);
 }
+
+/// <summary>
+/// Creates or upgrades the API-key store schema to the version this build supports. Each store
+/// provider (SQLite, SQL Server) supplies one; hosts run it at startup and before admin commands.
+/// Implementations must be idempotent and safe to run concurrently from several nodes.
+/// </summary>
+public interface IApiKeyStoreMigrator
+{
+    /// <summary>Migrates the store schema.</summary>
+    /// <param name="ct">A token to cancel the operation.</param>
+    /// <returns>A task that completes when the schema is current.</returns>
+    Task MigrateAsync(CancellationToken ct);
+}
@@ -4,6 +4,20 @@ namespace ZB.MOM.WW.Auth.Abstractions.Ldap;

 public enum LdapTransport { Ldaps, StartTls, None }

+/// <summary>How <see cref="LdapOptions"/> authenticates a sign-in.</summary>
+public enum LdapBindMode
+{
+    /// <summary>Bind as the service account, search for the user, re-bind as the user (0.3.0 behaviour).</summary>
+    ServiceAccount,
+
+    /// <summary>
+    /// Bind as the signing-in user (identity built from <see cref="LdapOptions.UserBindTemplate"/>);
+    /// a successful bind authenticates them, and the same connection reads their own entry for groups.
+    /// No service account is used.
+    /// </summary>
+    User,
+}
+
 public sealed record LdapOptions
 {
    public bool Enabled { get; init; } = true;
@@ -19,6 +33,44 @@ public sealed record LdapOptions
    public string GroupAttribute { get; init; } = "memberOf";
    public int ConnectionTimeoutMs { get; init; } = 10_000;

+    /// <summary>
+    /// Selects the authentication flow. <see cref="LdapBindMode.ServiceAccount"/> (the default) is the
+    /// 0.3.0 bind-then-search-then-bind flow; <see cref="LdapBindMode.User"/> binds as the user and needs
+    /// no <see cref="ServiceAccountDn"/> / <see cref="ServiceAccountPassword"/>.
+    /// </summary>
+    public LdapBindMode BindMode { get; init; } = LdapBindMode.ServiceAccount;
+
+    /// <summary>
+    /// <see cref="LdapBindMode.User"/> only: the simple-bind identity, with <c>{0}</c> replaced by the
+    /// normalised login name. Must contain <c>{0}</c> exactly once.
+    /// <para>
+    /// Prefer the down-level form <c>EXAMPLE\{0}</c>: it maps exactly to the <c>sAMAccountName</c> the
+    /// group search uses, so the account that binds is the account whose groups are read. A UPN
+    /// template (<c>{0}@example.com</c>) is only correct where every account's UPN prefix equals its
+    /// <c>sAMAccountName</c> — AD resolves a UPN bind against the explicit <c>userPrincipalName</c>, so
+    /// a mismatch can fail sign-in or bind a different account than the one whose groups are read.
+    /// </para>
+    /// </summary>
+    public string UserBindTemplate { get; init; } = "";
+
+    /// <summary>
+    /// <see cref="LdapBindMode.User"/> only: when <see langword="true"/> (default) a login typed as
+    /// <c>DOMAIN\name</c> or <c>name@suffix</c> is reduced to <c>name</c>; when <see langword="false"/>
+    /// such input is refused as bad credentials without contacting the directory.
+    /// </summary>
+    public bool AcceptDomainQualifiedNames { get; init; } = true;
+
+    /// <summary>
+    /// Ordered fallback LDAP endpoints tried when the primary (and any earlier fallback)
+    /// fails with a system-side error (connect/TLS, service-account bind, or search).
+    /// Each entry is <c>"host"</c> (adopts <see cref="Port"/>) or <c>"host:port"</c>.
+    /// Empty (the default) disables failover entirely — single-endpoint behavior is
+    /// unchanged. The TLS validation policy (<see cref="ServerCertificateValidationCallback"/>
+    /// / <see cref="AllowInsecure"/>) applies to every endpoint. Endpoint preference is
+    /// sticky: the service keeps using the last endpoint that answered until it fails.
+    /// </summary>
+    public IReadOnlyList<string> FallbackServers { get; init; } = Array.Empty<string>();
+
    /// <summary>
    /// Optional hook to harden (or, in dev, relax) TLS server-certificate validation for the
    /// <see cref="LdapTransport.Ldaps"/> / <see cref="LdapTransport.StartTls"/> transports. When
@@ -0,0 +1,64 @@
+using Microsoft.Data.SqlClient;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer;
+
+/// <summary>
+/// Opens connections to the shared SQL Server API-key database and carries the settings every
+/// command needs (schema name, command timeout).
+/// </summary>
+/// <param name="options">The SQL Server API-key store options; validated in the constructor.</param>
+/// <exception cref="InvalidOperationException"><paramref name="options"/> is not usable.</exception>
+public sealed class AuthSqlServerConnectionFactory(SqlServerApiKeyOptions options)
+{
+    private readonly SqlServerApiKeyOptions _options = Validate(options);
+
+    private static SqlServerApiKeyOptions Validate(SqlServerApiKeyOptions options)
+    {
+        ArgumentNullException.ThrowIfNull(options);
+        options.Validate();
+        return options;
+    }
+
+    /// <summary>The schema that owns the three API-key tables.</summary>
+    public string Schema => _options.Schema;
+
+    /// <summary>Command timeout, in whole seconds, for every command issued against this store.</summary>
+    public int CommandTimeoutSeconds => Math.Max(1, (int)_options.CommandTimeout.TotalSeconds);
+
+    /// <summary>Opens a connection to the shared API-key database.</summary>
+    /// <param name="cancellationToken">A token to cancel the connect.</param>
+    /// <returns>An opened connection. The caller owns disposal.</returns>
+    public async Task<SqlConnection> OpenConnectionAsync(CancellationToken cancellationToken)
+    {
+        var connection = new SqlConnection(_options.ConnectionString);
+        try
+        {
+            await connection.OpenAsync(cancellationToken).ConfigureAwait(false);
+            return connection;
+        }
+        catch
+        {
+            await connection.DisposeAsync().ConfigureAwait(false);
+            throw;
+        }
+    }
+
+    /// <summary>
+    /// Creates a command on <paramref name="connection"/> with the configured timeout applied.
+    /// </summary>
+    /// <param name="connection">The open connection.</param>
+    /// <param name="commandText">The command text.</param>
+    /// <param name="transaction">The ambient transaction, if any.</param>
+    /// <returns>The configured command. The caller owns disposal.</returns>
+    public SqlCommand CreateCommand(
+        SqlConnection connection, string commandText, SqlTransaction? transaction = null)
+    {
+        ArgumentNullException.ThrowIfNull(connection);
+
+        SqlCommand command = connection.CreateCommand();
+        command.CommandText = commandText;
+        command.CommandTimeout = CommandTimeoutSeconds;
+        command.Transaction = transaction;
+        return command;
+    }
+}
@@ -0,0 +1,96 @@
+using Microsoft.Extensions.Configuration;
+using Microsoft.Extensions.DependencyInjection;
+using Microsoft.Extensions.DependencyInjection.Extensions;
+using Microsoft.Extensions.Options;
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+using ZB.MOM.WW.Auth.ApiKeys.DependencyInjection;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer.DependencyInjection;
+
+/// <summary>
+/// Dependency-injection entry point that swaps the API-key store backend from the SQLite default
+/// to a shared SQL Server database — one row set read and written by every node of a cluster (e.g.
+/// a ScadaBridge central pair), instead of the per-node file <see cref="AddZbApiKeyAuth"/> alone
+/// would give each node.
+/// </summary>
+public static class SqlServerApiKeyServiceCollectionExtensions
+{
+    /// <summary>
+    /// Registers the SQL Server-backed <see cref="IApiKeyStore"/>, <see cref="IApiKeyAdminStore"/>,
+    /// <see cref="IApiKeyAuditStore"/> and <see cref="IApiKeyStoreMigrator"/>, then calls
+    /// <see cref="ApiKeyServiceCollectionExtensions.AddZbApiKeyAuth"/> to complete the registration
+    /// (verifier, pepper provider, migration hosted service).
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// <b>Call this INSTEAD of <see cref="ApiKeyServiceCollectionExtensions.AddZbApiKeyAuth"/></b>,
+    /// never both — this method calls it internally. The SQL Server stores and migrator are
+    /// registered here, BEFORE that inner call, precisely because
+    /// <see cref="ApiKeyServiceCollectionExtensions.AddZbApiKeyAuth"/> registers everything with
+    /// <c>TryAdd*</c>: whichever implementation of <see cref="IApiKeyStore"/> (etc.) is registered
+    /// first wins, and a later <c>TryAdd</c> for the same service is a no-op. That ordering is the
+    /// entire mechanism by which this method displaces the SQLite default.
+    /// </para>
+    /// <para>
+    /// <see cref="ApiKeyServiceCollectionExtensions.AddZbApiKeyAuth"/> still registers a SQLite
+    /// connection factory from <c>ApiKeyOptions.SqlitePath</c> even when SQL Server wins every
+    /// store slot. That is harmless — the SQLite factory is lazy and the connection it would open
+    /// is never opened when nothing resolves it — and it is actually useful: it keeps the one path
+    /// a node's pre-existing per-node SQLite file lives at in a single option, which is exactly the
+    /// source path <see cref="SqliteApiKeyStoreImporter"/> needs for the one-shot copy into the
+    /// shared store.
+    /// </para>
+    /// <para>
+    /// <see cref="SqliteApiKeyStoreImporter"/> is registered here so a host can resolve it and run
+    /// it once at boot (e.g. from a startup task or its own hosted service) to copy an existing
+    /// node's per-node keys and audit rows into the now-shared store; it performs no work by being
+    /// merely registered.
+    /// </para>
+    /// <para>
+    /// <b>The connection string must be a literal value or an environment-variable override — never
+    /// a <c>${secret:}</c> reference</b> on a host whose secret store is itself SQL-backed: resolving
+    /// that reference would require a working secret store, which on such a host requires this very
+    /// database to already be reachable, an unbreakable circular dependency at boot.
+    /// </para>
+    /// </remarks>
+    /// <param name="services">The service collection to add to.</param>
+    /// <param name="config">The application configuration.</param>
+    /// <param name="apiKeySectionPath">Path of the configuration section holding the core API-key options.</param>
+    /// <param name="sqlServerSectionPath">Path of the configuration section holding <see cref="SqlServerApiKeyOptions"/>.</param>
+    /// <returns>The same <paramref name="services"/> instance, for chaining.</returns>
+    public static IServiceCollection AddZbApiKeyAuthSqlServer(
+        this IServiceCollection services,
+        IConfiguration config,
+        string apiKeySectionPath,
+        string sqlServerSectionPath)
+    {
+        ArgumentNullException.ThrowIfNull(services);
+        ArgumentNullException.ThrowIfNull(config);
+        ArgumentException.ThrowIfNullOrWhiteSpace(apiKeySectionPath);
+        ArgumentException.ThrowIfNullOrWhiteSpace(sqlServerSectionPath);
+
+        services.Configure<SqlServerApiKeyOptions>(config.GetSection(sqlServerSectionPath));
+
+        services.TryAddSingleton(sp =>
+            new AuthSqlServerConnectionFactory(sp.GetRequiredService<IOptions<SqlServerApiKeyOptions>>().Value));
+
+        // Registered BEFORE AddZbApiKeyAuth so its TryAdd calls find these already present and
+        // leave them alone — that is the whole mechanism by which the SQL Server store displaces
+        // SQLite (see the class remarks).
+        services.TryAddSingleton<IApiKeyStore>(sp =>
+            new SqlServerApiKeyStore(sp.GetRequiredService<AuthSqlServerConnectionFactory>()));
+        services.TryAddSingleton<IApiKeyAdminStore>(sp =>
+            new SqlServerApiKeyAdminStore(sp.GetRequiredService<AuthSqlServerConnectionFactory>()));
+        services.TryAddSingleton<IApiKeyAuditStore>(sp =>
+            new SqlServerApiKeyAuditStore(sp.GetRequiredService<AuthSqlServerConnectionFactory>()));
+
+        services.TryAddSingleton<SqlServerApiKeyStoreMigrator>();
+        services.TryAddSingleton<IApiKeyStoreMigrator>(sp => sp.GetRequiredService<SqlServerApiKeyStoreMigrator>());
+
+        // Registered so a host can resolve it and run the one-shot per-node SQLite -> shared store
+        // copy at boot; registering it here does no work on its own.
+        services.TryAddSingleton<SqliteApiKeyStoreImporter>();
+
+        return services.AddZbApiKeyAuth(config, apiKeySectionPath);
+    }
+}
@@ -0,0 +1,247 @@
+using System.Data;
+using Microsoft.Data.SqlClient;
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+using ZB.MOM.WW.Auth.ApiKeys.Sqlite;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer;
+
+/// <summary>
+/// SQL Server-backed administration store for API keys (create, revoke, rotate, delete,
+/// set-scopes, enable/disable, list).
+/// </summary>
+/// <remarks>
+/// Semantics are a deliberate one-for-one port of <c>SqliteApiKeyAdminStore</c> — same WHERE and SET
+/// clauses, same rows-affected return contract — so an admin command behaves identically whichever
+/// provider a host composed. The point of this provider is only WHERE the rows live: one shared row
+/// set for a clustered pair instead of a file per node, so a revoke on either node is a revoke
+/// everywhere.
+/// </remarks>
+/// <param name="connectionFactory">Factory for connections to the shared API-key database.</param>
+public sealed class SqlServerApiKeyAdminStore(AuthSqlServerConnectionFactory connectionFactory) : IApiKeyAdminStore
+{
+    /// <inheritdoc />
+    /// <remarks>
+    /// A plain INSERT: a duplicate <c>key_id</c> surfaces as a primary-key
+    /// <see cref="SqlException"/> rather than silently overwriting an existing key (the SQLite store
+    /// raises its own <c>SqliteException</c> in the same situation).
+    /// </remarks>
+    public async Task CreateAsync(ApiKeyRecord record, CancellationToken ct)
+    {
+        ArgumentNullException.ThrowIfNull(record);
+
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        await using SqlCommand command = connectionFactory.CreateCommand(connection, $"""
+            INSERT INTO [{schema}].[{SqlServerAuthSchema.ApiKeysTable}] (
+                key_id, key_prefix, secret_hash, display_name, scopes,
+                constraints, created_utc, last_used_utc, revoked_utc, expires_utc)
+            VALUES (
+                @key_id, @key_prefix, @secret_hash, @display_name, @scopes,
+                @constraints, @created_utc, @last_used_utc, @revoked_utc, @expires_utc);
+            """);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = record.KeyId;
+        command.Parameters.Add("@key_prefix", SqlDbType.NVarChar, 16).Value = record.KeyPrefix;
+        command.Parameters.Add("@secret_hash", SqlDbType.VarBinary, 64).Value = record.SecretHash;
+        command.Parameters.Add("@display_name", SqlDbType.NVarChar, 256).Value = record.DisplayName;
+        command.Parameters.Add("@scopes", SqlDbType.NVarChar, -1).Value = ScopeSerializer.Serialize(record.Scopes);
+        command.Parameters.Add("@constraints", SqlDbType.NVarChar, -1).Value = (object?)record.ConstraintsJson ?? DBNull.Value;
+        command.Parameters.Add("@created_utc", SqlDbType.DateTimeOffset).Value = record.CreatedUtc;
+        command.Parameters.Add("@last_used_utc", SqlDbType.DateTimeOffset).Value = (object?)record.LastUsedUtc ?? DBNull.Value;
+        command.Parameters.Add("@revoked_utc", SqlDbType.DateTimeOffset).Value = (object?)record.RevokedUtc ?? DBNull.Value;
+        command.Parameters.Add("@expires_utc", SqlDbType.DateTimeOffset).Value = (object?)record.ExpiresUtc ?? DBNull.Value;
+
+        await command.ExecuteNonQueryAsync(ct).ConfigureAwait(false);
+    }
+
+    /// <inheritdoc />
+    /// <remarks>
+    /// Guarded on <c>revoked_utc IS NULL</c>, so the returned flag means "this call revoked it" —
+    /// a second revoke of the same key returns <see langword="false"/> and preserves the original
+    /// revocation instant.
+    /// </remarks>
+    public async Task<bool> RevokeAsync(string keyId, DateTimeOffset whenUtc, CancellationToken ct)
+    {
+        ArgumentException.ThrowIfNullOrWhiteSpace(keyId);
+
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        await using SqlCommand command = connectionFactory.CreateCommand(connection, $"""
+            UPDATE [{schema}].[{SqlServerAuthSchema.ApiKeysTable}]
+            SET revoked_utc = @revoked_utc
+            WHERE key_id = @key_id AND revoked_utc IS NULL;
+            """);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = keyId;
+        command.Parameters.Add("@revoked_utc", SqlDbType.DateTimeOffset).Value = whenUtc;
+
+        int rows = await command.ExecuteNonQueryAsync(ct).ConfigureAwait(false);
+
+        return rows > 0;
+    }
+
+    /// <inheritdoc />
+    /// <remarks>
+    /// Rotating also clears <c>last_used_utc</c> and <c>revoked_utc</c> — a rotated key is a fresh,
+    /// active credential — while leaving <c>expires_utc</c> alone, so an absolute expiry survives a
+    /// rotation.
+    /// </remarks>
+    public async Task<bool> RotateAsync(string keyId, byte[] newSecretHash, CancellationToken ct)
+    {
+        ArgumentException.ThrowIfNullOrWhiteSpace(keyId);
+        ArgumentNullException.ThrowIfNull(newSecretHash);
+
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        await using SqlCommand command = connectionFactory.CreateCommand(connection, $"""
+            UPDATE [{schema}].[{SqlServerAuthSchema.ApiKeysTable}]
+            SET secret_hash = @secret_hash,
+                last_used_utc = NULL,
+                revoked_utc = NULL
+            WHERE key_id = @key_id;
+            """);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = keyId;
+        command.Parameters.Add("@secret_hash", SqlDbType.VarBinary, 64).Value = newSecretHash;
+
+        int rows = await command.ExecuteNonQueryAsync(ct).ConfigureAwait(false);
+
+        return rows > 0;
+    }
+
+    /// <inheritdoc />
+    public async Task<bool> SetScopesAsync(string keyId, IReadOnlySet<string> scopes, CancellationToken ct)
+    {
+        ArgumentException.ThrowIfNullOrWhiteSpace(keyId);
+        ArgumentNullException.ThrowIfNull(scopes);
+
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        await using SqlCommand command = connectionFactory.CreateCommand(connection, $"""
+            UPDATE [{schema}].[{SqlServerAuthSchema.ApiKeysTable}]
+            SET scopes = @scopes
+            WHERE key_id = @key_id;
+            """);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = keyId;
+        command.Parameters.Add("@scopes", SqlDbType.NVarChar, -1).Value = ScopeSerializer.Serialize(scopes);
+
+        int rows = await command.ExecuteNonQueryAsync(ct).ConfigureAwait(false);
+
+        return rows > 0;
+    }
+
+    /// <inheritdoc />
+    public async Task<bool> SetEnabledAsync(string keyId, bool enabled, DateTimeOffset whenUtc, CancellationToken ct)
+    {
+        ArgumentException.ThrowIfNullOrWhiteSpace(keyId);
+
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        // Reversible toggle: NO `revoked_utc IS NULL` guard (unlike RevokeAsync), so it works
+        // regardless of current state. Deliberately leaves secret_hash and last_used_utc untouched
+        // — that is what distinguishes re-enable from RotateAsync.
+        SqlCommand command;
+
+        if (enabled)
+        {
+            command = connectionFactory.CreateCommand(connection, $"""
+                UPDATE [{schema}].[{SqlServerAuthSchema.ApiKeysTable}]
+                SET revoked_utc = NULL
+                WHERE key_id = @key_id;
+                """);
+            command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = keyId;
+        }
+        else
+        {
+            command = connectionFactory.CreateCommand(connection, $"""
+                UPDATE [{schema}].[{SqlServerAuthSchema.ApiKeysTable}]
+                SET revoked_utc = @revoked_utc
+                WHERE key_id = @key_id;
+                """);
+            command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = keyId;
+            command.Parameters.Add("@revoked_utc", SqlDbType.DateTimeOffset).Value = whenUtc;
+        }
+
+        await using (command)
+        {
+            int rows = await command.ExecuteNonQueryAsync(ct).ConfigureAwait(false);
+
+            return rows > 0;
+        }
+    }
+
+    /// <inheritdoc />
+    /// <remarks>
+    /// Only a REVOKED key can be deleted (<c>revoked_utc IS NOT NULL</c>), so an active credential
+    /// cannot vanish in one step — revoke first, and the revocation stands on its own until then.
+    /// </remarks>
+    public async Task<bool> DeleteAsync(string keyId, CancellationToken ct)
+    {
+        ArgumentException.ThrowIfNullOrWhiteSpace(keyId);
+
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        await using SqlCommand command = connectionFactory.CreateCommand(connection, $"""
+            DELETE FROM [{schema}].[{SqlServerAuthSchema.ApiKeysTable}]
+            WHERE key_id = @key_id AND revoked_utc IS NOT NULL;
+            """);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = keyId;
+
+        int rows = await command.ExecuteNonQueryAsync(ct).ConfigureAwait(false);
+
+        return rows > 0;
+    }
+
+    /// <inheritdoc />
+    public async Task<IReadOnlyList<ApiKeyListItem>> ListAsync(CancellationToken ct)
+    {
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        // Deliberately omits secret_hash so listing can never leak secret material. key_id is the
+        // tie-break so keys minted in the same instant still list in a stable order.
+        await using SqlCommand command = connectionFactory.CreateCommand(connection, $"""
+            SELECT key_id, key_prefix, display_name, scopes, constraints,
+                   created_utc, last_used_utc, revoked_utc, expires_utc
+            FROM [{schema}].[{SqlServerAuthSchema.ApiKeysTable}]
+            ORDER BY created_utc DESC, key_id DESC;
+            """);
+
+        List<ApiKeyListItem> items = [];
+
+        await using SqlDataReader reader = await command.ExecuteReaderAsync(ct).ConfigureAwait(false);
+
+        while (await reader.ReadAsync(ct).ConfigureAwait(false))
+        {
+            items.Add(new ApiKeyListItem(
+                KeyId: reader.GetString(0),
+                KeyPrefix: reader.GetString(1),
+                DisplayName: reader.GetString(2),
+                Scopes: ScopeSerializer.Deserialize(reader.GetString(3)),
+                ConstraintsJson: reader.IsDBNull(4) ? null : reader.GetString(4),
+                CreatedUtc: reader.GetDateTimeOffset(5),
+                LastUsedUtc: reader.IsDBNull(6) ? null : reader.GetDateTimeOffset(6),
+                RevokedUtc: reader.IsDBNull(7) ? null : reader.GetDateTimeOffset(7),
+                ExpiresUtc: reader.IsDBNull(8) ? null : reader.GetDateTimeOffset(8)));
+        }
+
+        return items;
+    }
+}
@@ -0,0 +1,80 @@
+using System.Data;
+using Microsoft.Data.SqlClient;
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer;
+
+/// <summary>
+/// SQL Server-backed, append-only audit store for API-key events. A one-for-one port of
+/// <c>SqliteApiKeyAuditStore</c>; on a clustered host the rows from every node land in one table,
+/// so the audit trail no longer has to be stitched together per node.
+/// </summary>
+/// <param name="connectionFactory">Factory for connections to the shared API-key database.</param>
+public sealed class SqlServerApiKeyAuditStore(AuthSqlServerConnectionFactory connectionFactory) : IApiKeyAuditStore
+{
+    /// <inheritdoc />
+    public async Task AppendAsync(ApiKeyAuditEntry entry, CancellationToken ct)
+    {
+        ArgumentNullException.ThrowIfNull(entry);
+
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        await using SqlCommand command = connectionFactory.CreateCommand(connection, $"""
+            INSERT INTO [{schema}].[{SqlServerAuthSchema.ApiKeyAuditTable}] (
+                key_id, event_type, remote_address, created_utc, details)
+            VALUES (@key_id, @event_type, @remote_address, @created_utc, @details);
+            """);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = (object?)entry.KeyId ?? DBNull.Value;
+        command.Parameters.Add("@event_type", SqlDbType.NVarChar, 64).Value = entry.EventType;
+        command.Parameters.Add("@remote_address", SqlDbType.NVarChar, 64).Value = (object?)entry.RemoteAddress ?? DBNull.Value;
+        command.Parameters.Add("@created_utc", SqlDbType.DateTimeOffset).Value = entry.CreatedUtc;
+        command.Parameters.Add("@details", SqlDbType.NVarChar, -1).Value = (object?)entry.Details ?? DBNull.Value;
+
+        await command.ExecuteNonQueryAsync(ct).ConfigureAwait(false);
+    }
+
+    /// <inheritdoc />
+    /// <remarks>
+    /// Ordered by the IDENTITY <c>audit_id</c>, not by <c>created_utc</c>: two events stamped in the
+    /// same instant still come back in insertion order. A non-positive limit short-circuits without
+    /// opening a connection — <c>SELECT TOP (-1)</c> is not even legal T-SQL.
+    /// </remarks>
+    public async Task<IReadOnlyList<ApiKeyAuditEntry>> ListRecentAsync(int limit, CancellationToken ct)
+    {
+        if (limit <= 0)
+        {
+            return [];
+        }
+
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        await using SqlCommand command = connectionFactory.CreateCommand(connection, $"""
+            SELECT TOP (@limit) key_id, event_type, remote_address, created_utc, details
+            FROM [{schema}].[{SqlServerAuthSchema.ApiKeyAuditTable}]
+            ORDER BY audit_id DESC;
+            """);
+        command.Parameters.Add("@limit", SqlDbType.Int).Value = limit;
+
+        List<ApiKeyAuditEntry> entries = [];
+
+        await using SqlDataReader reader = await command.ExecuteReaderAsync(ct).ConfigureAwait(false);
+
+        while (await reader.ReadAsync(ct).ConfigureAwait(false))
+        {
+            entries.Add(new ApiKeyAuditEntry(
+                KeyId: reader.IsDBNull(0) ? null : reader.GetString(0),
+                EventType: reader.GetString(1),
+                RemoteAddress: reader.IsDBNull(2) ? null : reader.GetString(2),
+                CreatedUtc: reader.GetDateTimeOffset(3),
+                Details: reader.IsDBNull(4) ? null : reader.GetString(4)));
+        }
+
+        return entries;
+    }
+}
@@ -0,0 +1,65 @@
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer;
+
+/// <summary>
+/// Options for the shared SQL Server API-key store (section supplied by the host, e.g.
+/// <c>ScadaBridge:InboundApi:ApiKeyStore:SqlServer</c>).
+/// </summary>
+public sealed class SqlServerApiKeyOptions
+{
+    /// <summary>
+    /// Full SQL Server connection string. Required. Never a <c>${secret:}</c> reference on hosts
+    /// whose secret store is itself SQL-backed.
+    /// </summary>
+    public string ConnectionString { get; set; } = string.Empty;
+
+    /// <summary>Schema that owns the three tables. Letters, digits and underscore only (it is interpolated into DDL).</summary>
+    public string Schema { get; set; } = "zbauth";
+
+    /// <summary>
+    /// Command timeout applied to every store operation. Defaults to 30 seconds — long enough for
+    /// a contended shared database, short enough that a hung server surfaces as an error rather
+    /// than stalling a resolve indefinitely.
+    /// </summary>
+    public TimeSpan CommandTimeout { get; set; } = TimeSpan.FromSeconds(30);
+
+    /// <summary>Fails fast on a blank connection string, an unsafe schema identifier, or a non-positive command timeout.</summary>
+    /// <exception cref="InvalidOperationException">The options are not usable.</exception>
+    public void Validate()
+    {
+        if (string.IsNullOrWhiteSpace(ConnectionString))
+            throw new InvalidOperationException("SQL Server API-key store requires a ConnectionString.");
+
+        ValidateIdentifier(Schema);
+
+        if (CommandTimeout <= TimeSpan.Zero)
+            throw new InvalidOperationException($"SQL Server API-key store CommandTimeout must be positive (was {CommandTimeout}).");
+    }
+
+    /// <summary>
+    /// Strict allow-list for a SQL identifier that will be interpolated into DDL or query text.
+    /// Rejects everything that is not <c>[A-Za-z_][A-Za-z0-9_]{0,63}</c> — no brackets, quotes, dots
+    /// or spaces — so no bracket-escape (<c>]]</c>) trick can break out of the quoting applied at the
+    /// use site. Uses the ASCII-only character tests deliberately, so homoglyph and full-width
+    /// look-alikes are rejected rather than accepted as "letters".
+    /// </summary>
+    /// <param name="schemaName">The candidate identifier.</param>
+    /// <returns>The same value, once validated.</returns>
+    /// <exception cref="InvalidOperationException">The value is not a plain identifier.</exception>
+    /// <remarks>
+    /// Public because <see cref="SqlServerAuthSchema.CreateSchemaDdl(string)"/> is public and must
+    /// defend itself rather than trust that every caller routed through <see cref="Validate"/>.
+    /// </remarks>
+    public static string ValidateIdentifier(string schemaName)
+    {
+        if (string.IsNullOrEmpty(schemaName)
+            || schemaName.Length > 64
+            || (!char.IsAsciiLetter(schemaName[0]) && schemaName[0] != '_')
+            || !schemaName.All(c => char.IsAsciiLetterOrDigit(c) || c == '_'))
+        {
+            throw new InvalidOperationException(
+                $"SQL Server API-key store schema name '{schemaName}' is not a safe identifier.");
+        }
+
+        return schemaName;
+    }
+}
@@ -0,0 +1,100 @@
+using System.Data;
+using Microsoft.Data.SqlClient;
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+using ZB.MOM.WW.Auth.ApiKeys.Sqlite;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer;
+
+/// <summary>
+/// SQL Server-backed read store for API-key records — the verification hot path.
+/// </summary>
+/// <remarks>
+/// Semantics are a deliberate one-for-one port of <c>SqliteApiKeyStore</c>, down to the WHERE
+/// clauses, so a host can switch providers without any behaviour changing. One connection per
+/// operation: the shared store is reached over the network and pooling, not a held connection, is
+/// what keeps a central pair's concurrent verifies cheap.
+/// </remarks>
+/// <param name="connectionFactory">Factory for connections to the shared API-key database.</param>
+public sealed class SqlServerApiKeyStore(AuthSqlServerConnectionFactory connectionFactory) : IApiKeyStore
+{
+    private const string SelectColumns =
+        "key_id, key_prefix, secret_hash, display_name, scopes, constraints, created_utc, last_used_utc, revoked_utc, expires_utc";
+
+    /// <inheritdoc />
+    public Task<ApiKeyRecord?> FindByKeyIdAsync(string keyId, CancellationToken ct)
+    {
+        ArgumentException.ThrowIfNullOrWhiteSpace(keyId);
+        return FindAsync(keyId, requireActive: false, ct);
+    }
+
+    /// <inheritdoc />
+    public Task<ApiKeyRecord?> FindActiveByKeyIdAsync(string keyId, CancellationToken ct)
+    {
+        ArgumentException.ThrowIfNullOrWhiteSpace(keyId);
+        return FindAsync(keyId, requireActive: true, ct);
+    }
+
+    /// <inheritdoc />
+    /// <remarks>
+    /// Guarded on <c>revoked_utc IS NULL</c>: a revoked key's last-used stamp is frozen at its last
+    /// legitimate use, so it never records the rejected attempts that follow.
+    /// </remarks>
+    public async Task MarkUsedAsync(string keyId, DateTimeOffset whenUtc, CancellationToken ct)
+    {
+        ArgumentException.ThrowIfNullOrWhiteSpace(keyId);
+
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        await using SqlCommand command = connectionFactory.CreateCommand(connection, $"""
+            UPDATE [{schema}].[{SqlServerAuthSchema.ApiKeysTable}]
+            SET last_used_utc = @last_used_utc
+            WHERE key_id = @key_id AND revoked_utc IS NULL;
+            """);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = keyId;
+        command.Parameters.Add("@last_used_utc", SqlDbType.DateTimeOffset).Value = whenUtc;
+
+        await command.ExecuteNonQueryAsync(ct).ConfigureAwait(false);
+    }
+
+    private async Task<ApiKeyRecord?> FindAsync(string keyId, bool requireActive, CancellationToken ct)
+    {
+        string schema = connectionFactory.Schema;
+        string table = $"[{schema}].[{SqlServerAuthSchema.ApiKeysTable}]";
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(ct).ConfigureAwait(false);
+
+        // Active means NOT REVOKED only. Expiry is deliberately not filtered here — the verifier
+        // owns it, so it can tell an expired key from an unknown one (ApiKeyFailure.KeyExpired).
+        await using SqlCommand command = connectionFactory.CreateCommand(
+            connection,
+            requireActive
+                ? $"SELECT {SelectColumns} FROM {table} WHERE key_id = @key_id AND revoked_utc IS NULL;"
+                : $"SELECT {SelectColumns} FROM {table} WHERE key_id = @key_id;");
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = keyId;
+
+        await using SqlDataReader reader = await command.ExecuteReaderAsync(ct).ConfigureAwait(false);
+
+        if (!await reader.ReadAsync(ct).ConfigureAwait(false))
+        {
+            return null;
+        }
+
+        return ReadRecord(reader);
+    }
+
+    internal static ApiKeyRecord ReadRecord(SqlDataReader reader) => new(
+        KeyId: reader.GetString(0),
+        KeyPrefix: reader.GetString(1),
+        SecretHash: reader.GetFieldValue<byte[]>(2),
+        DisplayName: reader.GetString(3),
+        Scopes: ScopeSerializer.Deserialize(reader.GetString(4)),
+        ConstraintsJson: reader.IsDBNull(5) ? null : reader.GetString(5),
+        CreatedUtc: reader.GetDateTimeOffset(6),
+        LastUsedUtc: reader.IsDBNull(7) ? null : reader.GetDateTimeOffset(7),
+        RevokedUtc: reader.IsDBNull(8) ? null : reader.GetDateTimeOffset(8),
+        ExpiresUtc: reader.IsDBNull(9) ? null : reader.GetDateTimeOffset(9));
+}
@@ -0,0 +1,207 @@
+using System.Data;
+using System.Globalization;
+using Microsoft.Data.SqlClient;
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+using ZB.MOM.WW.Auth.ApiKeys.Sqlite;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer;
+
+/// <summary>
+/// Creates the shared SQL Server API-key schema and records the applied version. Idempotent: safe
+/// to run on every node at every startup, including concurrently — lost provisioning races surface
+/// as retryable <see cref="SqlException"/>s (see <see cref="IsRetryableMigrationErrorNumber"/>) and
+/// the retry re-runs the <c>IF NOT EXISTS</c> path against the objects the winner just created.
+/// Refuses to run against a database whose recorded version is newer than this build supports.
+/// </summary>
+/// <param name="connectionFactory">Factory for connections to the shared API-key database.</param>
+public sealed class SqlServerApiKeyStoreMigrator(AuthSqlServerConnectionFactory connectionFactory)
+    : IApiKeyStoreMigrator
+{
+    // Every node runs this at every startup, so a virgin database being provisioned by N nodes at
+    // once is the NORMAL case, not a rare race — and an unretried failure aborts that node's whole
+    // host at startup. Two failure shapes come out of that race, and retrying is correct for both
+    // precisely because the migration is idempotent:
+    //
+    //   - Deadlock victim (1205): concurrent Serializable DDL against the system catalogs can pick
+    //     a victim even though the IF NOT EXISTS guards make the logic itself safe.
+    //   - Lost CREATE / lost INSERT (2714, 1913, 2627): Serializable range locks do NOT serialize
+    //     the metadata existence checks against another session's in-flight CREATE — observed live
+    //     on the sibling shared secret store (ScadaBridge central-pair first boot, 2026-08-07):
+    //     both nodes passed the sys.schemas IF NOT EXISTS probe, the loser's CREATE SCHEMA then
+    //     died with 2714 "There is already an object named 'zbsecrets'". The object the loser tried
+    //     to create now exists, which is the desired end state — the retry's IF NOT EXISTS pass
+    //     skips it and proceeds.
+    //
+    // The retry cap keeps this from masking a GENUINE conflict (e.g. an unrelated pre-existing
+    // object colliding with the schema name): that reproduces on every attempt, so after
+    // MaxMigrationRetries the real SqlException propagates and the node fails closed.
+    private const int MaxMigrationRetries = 5;
+
+    /// <summary>Deadlock victim.</summary>
+    private const int SqlDeadlockVictimErrorNumber = 1205;
+
+    /// <summary>"There is already an object named '%ls'" — lost CREATE SCHEMA / CREATE TABLE race.</summary>
+    private const int SqlObjectAlreadyExistsErrorNumber = 2714;
+
+    /// <summary>"An index or statistics with name '%ls' already exists" — lost CREATE INDEX race.</summary>
+    private const int SqlIndexAlreadyExistsErrorNumber = 1913;
+
+    /// <summary>Primary-key violation — lost version-row MERGE-insert race.</summary>
+    private const int SqlPrimaryKeyViolationErrorNumber = 2627;
+
+    /// <inheritdoc />
+    /// <exception cref="AuthStoreMigrationException">
+    /// The recorded schema version is newer than <see cref="SqlServerAuthSchema.CurrentVersion"/>.
+    /// </exception>
+    public async Task MigrateAsync(CancellationToken cancellationToken)
+    {
+        for (int attempt = 1; ; attempt++)
+        {
+            try
+            {
+                await MigrateOnceAsync(cancellationToken).ConfigureAwait(false);
+                return;
+            }
+            catch (SqlException ex)
+                when (IsRetryableMigrationError(ex) && attempt < MaxMigrationRetries)
+            {
+                await Task.Delay(BackoffFor(attempt), cancellationToken).ConfigureAwait(false);
+            }
+        }
+    }
+
+    /// <summary>
+    /// Back-off before retry attempt <paramref name="attempt"/> + 1: a growing base plus RANDOM
+    /// jitter. The jitter is the load-bearing half. The sibling secret-store migrator this is ported
+    /// from delays a flat <c>100 * attempt</c>, which is the SAME value for every victim of the same
+    /// round — so N deadlock victims wake together and re-collide, exactly the lockstep its comment
+    /// claims to prevent. Measured here against a live SQL Server with eight concurrent migrators on
+    /// a virgin schema, the unjittered form exhausted the five-attempt budget on repeated 1205s in
+    /// roughly 1 run in 60; with jitter it did not.
+    /// </summary>
+    /// <param name="attempt">The 1-based attempt that just failed.</param>
+    /// <returns>How long to wait before the next attempt.</returns>
+    private static TimeSpan BackoffFor(int attempt) =>
+        TimeSpan.FromMilliseconds((100 * attempt) + Random.Shared.Next(0, 100 * attempt));
+
+    /// <summary>
+    /// Whether any error carried by <paramref name="exception"/> is a retryable provisioning-race
+    /// outcome. Scans the whole <see cref="SqlException.Errors"/> collection rather than only
+    /// <see cref="SqlException.Number"/> because a failed DDL batch stacks follow-on errors (e.g.
+    /// 2714 arrives with 2759 "CREATE SCHEMA failed due to previous errors") and the retryable one
+    /// is not guaranteed to be first.
+    /// </summary>
+    /// <param name="exception">The exception the migration attempt failed with.</param>
+    /// <returns><see langword="true"/> when a retry may succeed.</returns>
+    internal static bool IsRetryableMigrationError(SqlException exception)
+    {
+        foreach (SqlError error in exception.Errors)
+        {
+            if (IsRetryableMigrationErrorNumber(error.Number))
+            {
+                return true;
+            }
+        }
+
+        return IsRetryableMigrationErrorNumber(exception.Number);
+    }
+
+    /// <summary>
+    /// Classifies a SQL Server error number as a retryable outcome of N nodes provisioning the same
+    /// (possibly virgin) database at once. Everything here is a race whose retry re-runs the
+    /// idempotent <c>IF NOT EXISTS</c> path; anything else (login failures, missing database,
+    /// permissions, syntax) reproduces identically on retry and must propagate. 2705 (duplicate
+    /// column) is deliberately absent: every column lives inside a single <c>CREATE TABLE</c>
+    /// statement, so no cross-session race can produce it.
+    /// </summary>
+    /// <param name="errorNumber">The <see cref="SqlError.Number"/> to classify.</param>
+    /// <returns><see langword="true"/> when the number is a known provisioning-race outcome.</returns>
+    internal static bool IsRetryableMigrationErrorNumber(int errorNumber) => errorNumber
+        is SqlDeadlockVictimErrorNumber
+        or SqlObjectAlreadyExistsErrorNumber
+        or SqlIndexAlreadyExistsErrorNumber
+        or SqlPrimaryKeyViolationErrorNumber;
+
+    private async Task MigrateOnceAsync(CancellationToken cancellationToken)
+    {
+        string schema = connectionFactory.Schema;
+
+        await using SqlConnection connection =
+            await connectionFactory.OpenConnectionAsync(cancellationToken).ConfigureAwait(false);
+
+        // Serializable: unlike the single-writer SQLite store, EVERY node in the cluster runs this
+        // migration at startup, so concurrent first-boots are the normal case rather than a rare
+        // race. The isolation level narrows the race window but does NOT close it — the IF NOT
+        // EXISTS metadata probes are not serialized against another session's in-flight CREATE, so
+        // a loser can still fail with "object exists" (2714/1913) or a deadlock (1205). The retry
+        // filter in MigrateAsync is what makes simultaneous provision converge: by the time the
+        // loser retries, the winner's objects exist and every guard here skips them.
+        await using SqlTransaction transaction = (SqlTransaction)
+            await connection.BeginTransactionAsync(IsolationLevel.Serializable, cancellationToken)
+                .ConfigureAwait(false);
+
+        int existingVersion =
+            await ReadExistingSchemaVersionAsync(connection, transaction, schema, cancellationToken)
+                .ConfigureAwait(false);
+
+        if (existingVersion > SqlServerAuthSchema.CurrentVersion)
+        {
+            throw new AuthStoreMigrationException(
+                $"Shared API-key database schema version {existingVersion} is newer than supported version " +
+                $"{SqlServerAuthSchema.CurrentVersion}. Upgrade this node before it writes to the shared store.");
+        }
+
+        await using (SqlCommand ddl = connectionFactory.CreateCommand(
+            connection, SqlServerAuthSchema.CreateSchemaDdl(schema), transaction))
+        {
+            await ddl.ExecuteNonQueryAsync(cancellationToken).ConfigureAwait(false);
+        }
+
+        // MERGE rather than insert-if-missing: WHEN MATCHED re-stamps the row, so a future schema
+        // version actually lands in the table instead of leaving the first-ever value behind.
+        await using (SqlCommand version = connectionFactory.CreateCommand(
+            connection,
+            $"""
+            MERGE [{schema}].[{SqlServerAuthSchema.SchemaVersionTable}] AS target
+            USING (SELECT 1 AS id, @version AS version, @applied_utc AS applied_utc) AS source
+                ON target.id = source.id
+            WHEN MATCHED THEN
+                UPDATE SET version = source.version, applied_utc = source.applied_utc
+            WHEN NOT MATCHED THEN
+                INSERT (id, version, applied_utc) VALUES (1, source.version, source.applied_utc);
+            """,
+            transaction))
+        {
+            version.Parameters.Add("@version", SqlDbType.Int).Value = SqlServerAuthSchema.CurrentVersion;
+            version.Parameters.Add("@applied_utc", SqlDbType.DateTimeOffset).Value = DateTimeOffset.UtcNow;
+            await version.ExecuteNonQueryAsync(cancellationToken).ConfigureAwait(false);
+        }
+
+        await transaction.CommitAsync(cancellationToken).ConfigureAwait(false);
+    }
+
+    private async Task<int> ReadExistingSchemaVersionAsync(
+        SqlConnection connection,
+        SqlTransaction transaction,
+        string schema,
+        CancellationToken cancellationToken)
+    {
+        // OBJECT_ID returns NULL for an absent table, so this reads 0 on a virgin database without
+        // needing the table to exist first.
+        await using SqlCommand command = connectionFactory.CreateCommand(
+            connection,
+            $"""
+            IF OBJECT_ID(N'[{schema}].[{SqlServerAuthSchema.SchemaVersionTable}]', N'U') IS NULL
+                SELECT 0;
+            ELSE
+                SELECT ISNULL((SELECT version FROM [{schema}].[{SqlServerAuthSchema.SchemaVersionTable}] WHERE id = 1), 0);
+            """,
+            transaction);
+
+        object? version = await command.ExecuteScalarAsync(cancellationToken).ConfigureAwait(false);
+
+        return version is null || version == DBNull.Value
+            ? 0
+            : Convert.ToInt32(version, CultureInfo.InvariantCulture);
+    }
+}
@@ -0,0 +1,116 @@
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer;
+
+/// <summary>
+/// Schema constants and DDL for the shared SQL Server API-key store. Mirrors
+/// <c>ZB.MOM.WW.Auth.ApiKeys.Sqlite.SqliteAuthSchema</c> column-for-column so a row means exactly
+/// the same thing in either store.
+/// </summary>
+/// <remarks>
+/// <para>
+/// <b>Timestamps are <c>DATETIMEOFFSET(7)</c>, not ISO-8601 text.</b> The sibling
+/// <c>ZB.MOM.WW.Secrets</c> SQL Server store deliberately stores its timestamps as
+/// <c>nvarchar</c> because its rows <em>replicate</em> with a SQLite store and last-writer-wins
+/// compares those timestamps for equality, where a sub-tick round-trip difference would turn a tie
+/// into a spurious "newer". No such rationale applies here: this store never replicates with the
+/// SQLite one — a host picks exactly one provider, and the one-shot SQLite import is a copy, not an
+/// ongoing sync — so the native type is used, which sorts and range-scans correctly and keeps the
+/// tables readable to a DBA.
+/// </para>
+/// <para>
+/// <c>SqliteAuthSchema.CurrentVersion</c> is 3 only because that store carries the donor
+/// MxAccessGateway's version history (v1/v2 were byte-identical; v3 added <c>expires_utc</c>). This
+/// schema is created whole, with <c>expires_utc</c> present from the start, so its own version line
+/// starts at 1.
+/// </para>
+/// </remarks>
+public static class SqlServerAuthSchema
+{
+    /// <summary>
+    /// The schema version this build creates and supports. The migrator refuses a database whose
+    /// recorded version is <em>newer</em> than this.
+    /// </summary>
+    public const int CurrentVersion = 1;
+
+    /// <summary>Default schema the API-key tables live in.</summary>
+    public const string DefaultSchemaName = "zbauth";
+
+    /// <summary>Name of the single-row table tracking the applied schema version.</summary>
+    public const string SchemaVersionTable = "schema_version";
+
+    /// <summary>Name of the table storing API-key records.</summary>
+    public const string ApiKeysTable = "api_keys";
+
+    /// <summary>Name of the append-only audit table.</summary>
+    public const string ApiKeyAuditTable = "api_key_audit";
+
+    /// <summary>
+    /// Returns the idempotent DDL that provisions the schema, the three tables and the two
+    /// supporting indexes in <paramref name="schemaName"/>.
+    /// </summary>
+    /// <param name="schemaName">The SQL schema name to create the objects in.</param>
+    /// <returns>The DDL batch.</returns>
+    /// <remarks>
+    /// <paramref name="schemaName"/> is interpolated, not parameterized — T-SQL has no parameter
+    /// form for an object name. It is re-validated here against the same strict identifier
+    /// allow-list <see cref="SqlServerApiKeyOptions.Validate"/> enforces, and bracket-quoted below.
+    /// The duplicate check is deliberate: this method is <c>public</c>, so it must defend itself
+    /// rather than trust that every caller routed through the options type. The version row is NOT
+    /// written here — <see cref="SqlServerApiKeyStoreMigrator"/> writes it with a MERGE that also
+    /// updates on match, so a future upgrade re-stamps the stored version.
+    /// </remarks>
+    /// <exception cref="InvalidOperationException"><paramref name="schemaName"/> is not a plain identifier.</exception>
+    public static string CreateSchemaDdl(string schemaName)
+    {
+        SqlServerApiKeyOptions.ValidateIdentifier(schemaName);
+
+        return $"""
+            IF NOT EXISTS (SELECT 1 FROM sys.schemas WHERE name = N'{schemaName}')
+                EXEC(N'CREATE SCHEMA [{schemaName}]');
+
+            IF OBJECT_ID(N'[{schemaName}].[{SchemaVersionTable}]', N'U') IS NULL
+                CREATE TABLE [{schemaName}].[{SchemaVersionTable}] (
+                    id          INT               NOT NULL CONSTRAINT PK_{schemaName}_{SchemaVersionTable} PRIMARY KEY CHECK (id = 1),
+                    version     INT               NOT NULL,
+                    applied_utc DATETIMEOFFSET(7) NOT NULL
+                );
+
+            IF OBJECT_ID(N'[{schemaName}].[{ApiKeysTable}]', N'U') IS NULL
+                CREATE TABLE [{schemaName}].[{ApiKeysTable}] (
+                    key_id        NVARCHAR(64)      NOT NULL CONSTRAINT PK_{schemaName}_{ApiKeysTable} PRIMARY KEY,
+                    key_prefix    NVARCHAR(16)      NOT NULL,
+                    secret_hash   VARBINARY(64)     NOT NULL,
+                    display_name  NVARCHAR(256)     NOT NULL,
+                    scopes        NVARCHAR(MAX)     NOT NULL,
+                    constraints   NVARCHAR(MAX)     NULL,
+                    created_utc   DATETIMEOFFSET(7) NOT NULL,
+                    last_used_utc DATETIMEOFFSET(7) NULL,
+                    revoked_utc   DATETIMEOFFSET(7) NULL,
+                    expires_utc   DATETIMEOFFSET(7) NULL
+                );
+
+            IF OBJECT_ID(N'[{schemaName}].[{ApiKeyAuditTable}]', N'U') IS NULL
+                CREATE TABLE [{schemaName}].[{ApiKeyAuditTable}] (
+                    audit_id       BIGINT IDENTITY(1,1) NOT NULL CONSTRAINT PK_{schemaName}_{ApiKeyAuditTable} PRIMARY KEY,
+                    key_id         NVARCHAR(64)      NULL,
+                    event_type     NVARCHAR(64)      NOT NULL,
+                    remote_address NVARCHAR(64)      NULL,
+                    created_utc    DATETIMEOFFSET(7) NOT NULL,
+                    details        NVARCHAR(MAX)     NULL
+                );
+
+            IF NOT EXISTS (
+                SELECT 1 FROM sys.indexes
+                WHERE name = N'ix_api_keys_revoked_utc'
+                  AND object_id = OBJECT_ID(N'[{schemaName}].[{ApiKeysTable}]'))
+                CREATE INDEX ix_api_keys_revoked_utc
+                    ON [{schemaName}].[{ApiKeysTable}] (revoked_utc);
+
+            IF NOT EXISTS (
+                SELECT 1 FROM sys.indexes
+                WHERE name = N'ix_api_key_audit_key_id_created_utc'
+                  AND object_id = OBJECT_ID(N'[{schemaName}].[{ApiKeyAuditTable}]'))
+                CREATE INDEX ix_api_key_audit_key_id_created_utc
+                    ON [{schemaName}].[{ApiKeyAuditTable}] (key_id, created_utc);
+            """;
+    }
+}
@@ -0,0 +1,300 @@
+using System.Data;
+using Microsoft.Data.SqlClient;
+using Microsoft.Data.Sqlite;
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+using ZB.MOM.WW.Auth.ApiKeys.Sqlite;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer;
+
+/// <summary>Result of a one-shot import.</summary>
+/// <param name="KeysInserted">Keys copied into SQL Server by this run.</param>
+/// <param name="KeysSkipped">Keys whose <c>key_id</c> already existed in SQL Server.</param>
+/// <param name="AuditInserted">Audit rows copied into SQL Server by this run.</param>
+/// <param name="AuditSkipped">Audit rows that already existed in SQL Server.</param>
+/// <param name="SourceFound">Whether the SQLite file existed at all.</param>
+public sealed record ApiKeyImportResult(
+    int KeysInserted, int KeysSkipped, int AuditInserted, int AuditSkipped, bool SourceFound);
+
+/// <summary>
+/// Copies every key and audit row from a per-node SQLite store into the shared SQL Server store,
+/// insert-where-not-exists (keys by key_id; audit rows by key_id + event_type + created_utc), so it is
+/// idempotent and every node of a pair may run it: the result is the UNION of the nodes' files. A row
+/// that already exists in SQL Server is never modified — a revoke recorded on one node but not the
+/// other therefore survives only if the revoked copy is imported first; operators reconcile by listing.
+/// </summary>
+/// <param name="sql">Factory for connections to the shared SQL Server API-key database.</param>
+public sealed class SqliteApiKeyStoreImporter(AuthSqlServerConnectionFactory sql)
+{
+    /// <summary>
+    /// Imports the keys and audit rows held in the SQLite store at <paramref name="sqlitePath"/>.
+    /// </summary>
+    /// <param name="sqlitePath">Path of the per-node SQLite API-key file.</param>
+    /// <param name="ct">A token to cancel the import.</param>
+    /// <returns>
+    /// The per-category insert/skip counts. When the file does not exist the result is all zeroes
+    /// with <see cref="ApiKeyImportResult.SourceFound"/> <see langword="false"/> and nothing is
+    /// opened — a host that has already imported (and renamed the file) pays nothing.
+    /// </returns>
+    /// <remarks>
+    /// <para>
+    /// The SQLite file is opened READ-ONLY and the schema migrator is deliberately NOT run against
+    /// it: the source is evidence, not something to upgrade. A file with no <c>api_keys</c> table is
+    /// treated as an empty source rather than an error, and a pre-v3 file (no <c>expires_utc</c>
+    /// column) reads its expiry as NULL = never expires, which is exactly what those keys meant.
+    /// </para>
+    /// <para>
+    /// Every row is read and the SQLite connection closed BEFORE any SQL Server write, and all the
+    /// writes then run inside ONE transaction. So the import is all-or-nothing: a failure part way
+    /// through rolls the transaction back and the exception propagates to the caller, leaving the
+    /// shared store exactly as it was. The host decides what that means — for a boot-time import,
+    /// logging and carrying on is usually right, because keys already in SQL Server keep working.
+    /// </para>
+    /// </remarks>
+    /// <exception cref="SqlException">A SQL Server write failed; nothing was committed.</exception>
+    /// <exception cref="SqliteException">The source file could not be read.</exception>
+    public async Task<ApiKeyImportResult> ImportAsync(string sqlitePath, CancellationToken ct)
+    {
+        ArgumentException.ThrowIfNullOrWhiteSpace(sqlitePath);
+
+        if (!File.Exists(sqlitePath))
+        {
+            return new ApiKeyImportResult(0, 0, 0, 0, SourceFound: false);
+        }
+
+        (IReadOnlyList<ApiKeyRecord> keys, IReadOnlyList<ApiKeyAuditEntry> auditEntries) =
+            await ReadSourceAsync(sqlitePath, ct).ConfigureAwait(false);
+
+        return await WriteAsync(keys, auditEntries, ct).ConfigureAwait(false);
+    }
+
+    // --- read side (SQLite) ---
+
+    private static async Task<(IReadOnlyList<ApiKeyRecord> Keys, IReadOnlyList<ApiKeyAuditEntry> Audit)>
+        ReadSourceAsync(string sqlitePath, CancellationToken ct)
+    {
+        SqliteConnectionStringBuilder builder = new()
+        {
+            DataSource = sqlitePath,
+            Mode = SqliteOpenMode.ReadOnly,
+        };
+
+        await using var connection = new SqliteConnection(builder.ToString());
+        await connection.OpenAsync(ct).ConfigureAwait(false);
+
+        IReadOnlyList<ApiKeyRecord> keys =
+            await TableExistsAsync(connection, SqliteAuthSchema.ApiKeysTable, ct).ConfigureAwait(false)
+                ? await ReadKeysAsync(connection, ct).ConfigureAwait(false)
+                : [];
+
+        IReadOnlyList<ApiKeyAuditEntry> audit =
+            await TableExistsAsync(connection, SqliteAuthSchema.ApiKeyAuditTable, ct).ConfigureAwait(false)
+                ? await ReadAuditAsync(connection, ct).ConfigureAwait(false)
+                : [];
+
+        return (keys, audit);
+    }
+
+    private static async Task<bool> TableExistsAsync(
+        SqliteConnection connection, string tableName, CancellationToken ct)
+    {
+        await using SqliteCommand command = connection.CreateCommand();
+        command.CommandText = """
+            SELECT COUNT(*) FROM sqlite_master WHERE type = 'table' AND name = $table_name;
+            """;
+        command.Parameters.AddWithValue("$table_name", tableName);
+
+        return (long)(await command.ExecuteScalarAsync(ct).ConfigureAwait(false) ?? 0L) > 0;
+    }
+
+    private static async Task<IReadOnlyList<ApiKeyRecord>> ReadKeysAsync(
+        SqliteConnection connection, CancellationToken ct)
+    {
+        // A pre-v3 file has no expires_utc column, so select a literal NULL in its place rather than
+        // failing the import: NULL is precisely what those keys mean (never expires).
+        bool hasExpires = await ColumnExistsAsync(
+            connection, SqliteAuthSchema.ApiKeysTable, SqliteAuthSchema.ExpiresUtcColumn, ct).ConfigureAwait(false);
+
+        string expiresColumn = hasExpires ? SqliteAuthSchema.ExpiresUtcColumn : "NULL";
+
+        await using SqliteCommand command = connection.CreateCommand();
+        command.CommandText =
+            "SELECT key_id, key_prefix, secret_hash, display_name, scopes, constraints, " +
+            $"created_utc, last_used_utc, revoked_utc, {expiresColumn} FROM api_keys;";
+
+        List<ApiKeyRecord> keys = [];
+
+        await using SqliteDataReader reader = await command.ExecuteReaderAsync(ct).ConfigureAwait(false);
+
+        while (await reader.ReadAsync(ct).ConfigureAwait(false))
+        {
+            keys.Add(new ApiKeyRecord(
+                KeyId: reader.GetString(0),
+                KeyPrefix: reader.GetString(1),
+                SecretHash: reader.GetFieldValue<byte[]>(2),
+                DisplayName: reader.GetString(3),
+                Scopes: ScopeSerializer.Deserialize(reader.GetString(4)),
+                ConstraintsJson: reader.IsDBNull(5) ? null : reader.GetString(5),
+                CreatedUtc: SqliteValueParsing.ParseUtc(reader.GetString(6)),
+                LastUsedUtc: SqliteValueParsing.ReadNullableUtc(reader, 7),
+                RevokedUtc: SqliteValueParsing.ReadNullableUtc(reader, 8),
+                ExpiresUtc: SqliteValueParsing.ReadNullableUtc(reader, 9)));
+        }
+
+        return keys;
+    }
+
+    // SQLite has no information_schema; PRAGMA table_info is the column probe (the same one
+    // SqliteAuthStoreMigrator uses to decide whether its v3 ALTER TABLE is needed).
+    private static async Task<bool> ColumnExistsAsync(
+        SqliteConnection connection, string tableName, string columnName, CancellationToken ct)
+    {
+        await using SqliteCommand command = connection.CreateCommand();
+        // table_info takes an identifier, not a bindable parameter; the name is an internal constant.
+        command.CommandText = $"PRAGMA table_info({tableName});";
+
+        await using SqliteDataReader reader = await command.ExecuteReaderAsync(ct).ConfigureAwait(false);
+
+        while (await reader.ReadAsync(ct).ConfigureAwait(false))
+        {
+            // PRAGMA table_info columns: cid(0), name(1), type(2), ...
+            if (string.Equals(reader.GetString(1), columnName, StringComparison.OrdinalIgnoreCase))
+            {
+                return true;
+            }
+        }
+
+        return false;
+    }
+
+    private static async Task<IReadOnlyList<ApiKeyAuditEntry>> ReadAuditAsync(
+        SqliteConnection connection, CancellationToken ct)
+    {
+        await using SqliteCommand command = connection.CreateCommand();
+        // Insertion order, so the shared table's IDENTITY re-creates the source's ordering.
+        command.CommandText = """
+            SELECT key_id, event_type, remote_address, created_utc, details
+            FROM api_key_audit
+            ORDER BY audit_id;
+            """;
+
+        List<ApiKeyAuditEntry> entries = [];
+
+        await using SqliteDataReader reader = await command.ExecuteReaderAsync(ct).ConfigureAwait(false);
+
+        while (await reader.ReadAsync(ct).ConfigureAwait(false))
+        {
+            entries.Add(new ApiKeyAuditEntry(
+                KeyId: reader.IsDBNull(0) ? null : reader.GetString(0),
+                EventType: reader.GetString(1),
+                RemoteAddress: reader.IsDBNull(2) ? null : reader.GetString(2),
+                CreatedUtc: SqliteValueParsing.ParseUtc(reader.GetString(3)),
+                Details: reader.IsDBNull(4) ? null : reader.GetString(4)));
+        }
+
+        return entries;
+    }
+
+    // --- write side (SQL Server) ---
+
+    private async Task<ApiKeyImportResult> WriteAsync(
+        IReadOnlyList<ApiKeyRecord> keys,
+        IReadOnlyList<ApiKeyAuditEntry> auditEntries,
+        CancellationToken ct)
+    {
+        string schema = sql.Schema;
+
+        await using SqlConnection connection = await sql.OpenConnectionAsync(ct).ConfigureAwait(false);
+        await using SqlTransaction transaction =
+            (SqlTransaction)await connection.BeginTransactionAsync(ct).ConfigureAwait(false);
+
+        int keysInserted = 0;
+        int auditInserted = 0;
+
+        foreach (ApiKeyRecord key in keys)
+        {
+            if (await InsertKeyIfAbsentAsync(connection, transaction, schema, key, ct).ConfigureAwait(false))
+            {
+                keysInserted++;
+            }
+        }
+
+        foreach (ApiKeyAuditEntry entry in auditEntries)
+        {
+            if (await InsertAuditIfAbsentAsync(connection, transaction, schema, entry, ct).ConfigureAwait(false))
+            {
+                auditInserted++;
+            }
+        }
+
+        await transaction.CommitAsync(ct).ConfigureAwait(false);
+
+        return new ApiKeyImportResult(
+            KeysInserted: keysInserted,
+            KeysSkipped: keys.Count - keysInserted,
+            AuditInserted: auditInserted,
+            AuditSkipped: auditEntries.Count - auditInserted,
+            SourceFound: true);
+    }
+
+    // Returns true when this call inserted the row. The guard is inside the statement rather than a
+    // separate SELECT so the check and the insert cannot straddle a concurrent writer.
+    private async Task<bool> InsertKeyIfAbsentAsync(
+        SqlConnection connection,
+        SqlTransaction transaction,
+        string schema,
+        ApiKeyRecord key,
+        CancellationToken ct)
+    {
+        await using SqlCommand command = sql.CreateCommand(connection, $"""
+            IF NOT EXISTS (
+                SELECT 1 FROM [{schema}].[{SqlServerAuthSchema.ApiKeysTable}] WHERE key_id = @key_id)
+                INSERT INTO [{schema}].[{SqlServerAuthSchema.ApiKeysTable}] (
+                    key_id, key_prefix, secret_hash, display_name, scopes,
+                    constraints, created_utc, last_used_utc, revoked_utc, expires_utc)
+                VALUES (
+                    @key_id, @key_prefix, @secret_hash, @display_name, @scopes,
+                    @constraints, @created_utc, @last_used_utc, @revoked_utc, @expires_utc);
+            """, transaction);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = key.KeyId;
+        command.Parameters.Add("@key_prefix", SqlDbType.NVarChar, 16).Value = key.KeyPrefix;
+        command.Parameters.Add("@secret_hash", SqlDbType.VarBinary, 64).Value = key.SecretHash;
+        command.Parameters.Add("@display_name", SqlDbType.NVarChar, 256).Value = key.DisplayName;
+        command.Parameters.Add("@scopes", SqlDbType.NVarChar, -1).Value = ScopeSerializer.Serialize(key.Scopes);
+        command.Parameters.Add("@constraints", SqlDbType.NVarChar, -1).Value = (object?)key.ConstraintsJson ?? DBNull.Value;
+        command.Parameters.Add("@created_utc", SqlDbType.DateTimeOffset).Value = key.CreatedUtc;
+        command.Parameters.Add("@last_used_utc", SqlDbType.DateTimeOffset).Value = (object?)key.LastUsedUtc ?? DBNull.Value;
+        command.Parameters.Add("@revoked_utc", SqlDbType.DateTimeOffset).Value = (object?)key.RevokedUtc ?? DBNull.Value;
+        command.Parameters.Add("@expires_utc", SqlDbType.DateTimeOffset).Value = (object?)key.ExpiresUtc ?? DBNull.Value;
+
+        return await command.ExecuteNonQueryAsync(ct).ConfigureAwait(false) > 0;
+    }
+
+    // Audit rows carry no natural key (audit_id is per-file IDENTITY), so identity is the triple
+    // key_id + event_type + created_utc. key_id is nullable, hence the explicit NULL-safe arm: a
+    // plain `key_id = @key_id` is UNKNOWN, never true, when both sides are NULL.
+    private async Task<bool> InsertAuditIfAbsentAsync(
+        SqlConnection connection,
+        SqlTransaction transaction,
+        string schema,
+        ApiKeyAuditEntry entry,
+        CancellationToken ct)
+    {
+        await using SqlCommand command = sql.CreateCommand(connection, $"""
+            IF NOT EXISTS (
+                SELECT 1 FROM [{schema}].[{SqlServerAuthSchema.ApiKeyAuditTable}]
+                WHERE ((key_id = @key_id) OR (key_id IS NULL AND @key_id IS NULL))
+                  AND event_type = @event_type
+                  AND created_utc = @created_utc)
+                INSERT INTO [{schema}].[{SqlServerAuthSchema.ApiKeyAuditTable}] (
+                    key_id, event_type, remote_address, created_utc, details)
+                VALUES (@key_id, @event_type, @remote_address, @created_utc, @details);
+            """, transaction);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = (object?)entry.KeyId ?? DBNull.Value;
+        command.Parameters.Add("@event_type", SqlDbType.NVarChar, 64).Value = entry.EventType;
+        command.Parameters.Add("@remote_address", SqlDbType.NVarChar, 64).Value = (object?)entry.RemoteAddress ?? DBNull.Value;
+        command.Parameters.Add("@created_utc", SqlDbType.DateTimeOffset).Value = entry.CreatedUtc;
+        command.Parameters.Add("@details", SqlDbType.NVarChar, -1).Value = (object?)entry.Details ?? DBNull.Value;
+
+        return await command.ExecuteNonQueryAsync(ct).ConfigureAwait(false) > 0;
+    }
+}
@@ -0,0 +1,42 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+  <ItemGroup>
+    <ProjectReference Include="..\ZB.MOM.WW.Auth.ApiKeys\ZB.MOM.WW.Auth.ApiKeys.csproj" />
+  </ItemGroup>
+
+  <ItemGroup>
+    <PackageReference Include="Microsoft.Data.SqlClient" />
+    <!--
+      Lightweight Microsoft.Extensions.* abstractions back the DI extension
+      (AddZbApiKeyAuthSqlServer, shipped by Task 6): IServiceCollection / TryAdd* / Configure
+      (DependencyInjection.Abstractions), IOptions (Options), IConfiguration
+      (Configuration.Abstractions), and IHostedService / AddHostedService (Hosting.Abstractions).
+    -->
+    <PackageReference Include="Microsoft.Extensions.DependencyInjection.Abstractions" />
+    <PackageReference Include="Microsoft.Extensions.Options" />
+    <!-- Supplies the Configure<TOptions>(IConfiguration) binding overload used by the DI extension. -->
+    <PackageReference Include="Microsoft.Extensions.Options.ConfigurationExtensions" />
+    <PackageReference Include="Microsoft.Extensions.Configuration.Abstractions" />
+    <PackageReference Include="Microsoft.Extensions.Hosting.Abstractions" />
+  </ItemGroup>
+
+  <PropertyGroup>
+    <TargetFramework>net10.0</TargetFramework>
+    <ImplicitUsings>enable</ImplicitUsings>
+    <Nullable>enable</Nullable>
+  </PropertyGroup>
+
+  <PropertyGroup>
+    <IsPackable>true</IsPackable>
+    <PackageId>ZB.MOM.WW.Auth.ApiKeys.SqlServer</PackageId>
+    <Authors>ZB.MOM.WW</Authors>
+    <Description>SQL Server API-key store for the ZB.MOM.WW SCADA family — one shared row set for clustered hosts (central pairs).</Description>
+    <PackageProjectUrl>https://gitea.dohertylan.com/dohertj2/zb-mom-ww-auth</PackageProjectUrl>
+    <RepositoryUrl>https://gitea.dohertylan.com/dohertj2/zb-mom-ww-auth</RepositoryUrl>
+  </PropertyGroup>
+
+  <ItemGroup>
+    <InternalsVisibleTo Include="ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests" />
+  </ItemGroup>
+
+</Project>
@@ -33,10 +33,15 @@ public sealed class ApiKeyAdminCommands
    private readonly IApiKeyAdminStore _adminStore;
    private readonly IApiKeyAuditStore _auditStore;
    private readonly IApiKeyPepperProvider _pepperProvider;
-    private readonly SqliteAuthStoreMigrator _migrator;
+    private readonly IApiKeyStoreMigrator _migrator;
    private readonly TimeProvider _clock;

-    /// <summary>Creates the command set over the supplied stores and options.</summary>
+    /// <summary>
+    /// Creates the command set over the supplied stores and options.
+    /// Retained for existing consumers (MxAccessGateway, HistorianGateway, ScadaBridge) that
+    /// construct this type against the concrete SQLite migrator; forwards to the
+    /// <see cref="IApiKeyStoreMigrator"/> overload.
+    /// </summary>
    /// <param name="options">API-key options (token prefix, store path, ...).</param>
    /// <param name="adminStore">Mutating store (create / revoke / rotate / delete / list).</param>
    /// <param name="auditStore">Append-only audit store wired into every mutating verb.</param>
@@ -50,6 +55,25 @@ public sealed class ApiKeyAdminCommands
        IApiKeyPepperProvider pepperProvider,
        SqliteAuthStoreMigrator migrator,
        TimeProvider? clock = null)
+        : this(options, adminStore, auditStore, pepperProvider, (IApiKeyStoreMigrator)migrator, clock)
+    {
+    }
+
+    /// <summary>Creates the command set over the supplied stores and options.</summary>
+    /// <param name="options">API-key options (token prefix, store path, ...).</param>
+    /// <param name="adminStore">Mutating store (create / revoke / rotate / delete / list).</param>
+    /// <param name="auditStore">Append-only audit store wired into every mutating verb.</param>
+    /// <param name="pepperProvider">Resolves the pepper used to hash secrets.</param>
+    /// <param name="migrator">Schema migrator used by <see cref="InitDbAsync"/> — whichever store
+    /// provider's implementation (SQLite, SQL Server, ...) DI has registered.</param>
+    /// <param name="clock">Optional clock; defaults to <see cref="TimeProvider.System"/>.</param>
+    public ApiKeyAdminCommands(
+        ApiKeyOptions options,
+        IApiKeyAdminStore adminStore,
+        IApiKeyAuditStore auditStore,
+        IApiKeyPepperProvider pepperProvider,
+        IApiKeyStoreMigrator migrator,
+        TimeProvider? clock = null)
    {
        ArgumentNullException.ThrowIfNull(options);
        ArgumentNullException.ThrowIfNull(adminStore);
@@ -1,17 +1,18 @@
 using Microsoft.Extensions.Hosting;
 using Microsoft.Extensions.Options;
 using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
-using ZB.MOM.WW.Auth.ApiKeys.Sqlite;

 namespace ZB.MOM.WW.Auth.ApiKeys.DependencyInjection;

 /// <summary>
-/// Runs the API-key SQLite schema migration at application startup when
-/// <see cref="ApiKeyOptions.RunMigrationsOnStartup"/> is <see langword="true"/>.
-/// The migration is idempotent, so repeated restarts are safe.
+/// Runs the registered <see cref="IApiKeyStoreMigrator"/> at application startup when
+/// <see cref="ApiKeyOptions.RunMigrationsOnStartup"/> is <see langword="true"/>. The migrator
+/// registered in DI is whichever store provider's implementation won the <c>TryAdd</c> race —
+/// SQLite by default, or a provider package (e.g. SQL Server) registered before
+/// <c>AddZbApiKeyAuth</c>. The migration is idempotent, so repeated restarts are safe.
 /// </summary>
 internal sealed class ApiKeyMigrationHostedService(
-    SqliteAuthStoreMigrator migrator,
+    IApiKeyStoreMigrator migrator,
    IOptions<ApiKeyOptions> options) : IHostedService
 {
    /// <inheritdoc />
@@ -61,10 +61,12 @@ public static class ApiKeyServiceCollectionExtensions
                sp.GetRequiredService<IApiKeyStore>(),
                sp.GetRequiredService<IApiKeyPepperProvider>()));

-        // Migrator: singleton, constructed from the already-registered connection factory.
-        // Needed before any store operations so the schema exists.
+        // Migrator: the SQLite implementation is the default; a provider package that registers
+        // IApiKeyStoreMigrator (and the three stores) BEFORE this call displaces it — that is the
+        // whole mechanism by which AddZbApiKeyAuthSqlServer swaps the backend.
        services.TryAddSingleton(sp =>
            new SqliteAuthStoreMigrator(sp.GetRequiredService<AuthSqliteConnectionFactory>()));
+        services.TryAddSingleton<IApiKeyStoreMigrator>(sp => sp.GetRequiredService<SqliteAuthStoreMigrator>());

        // Hosted service that runs migrations on startup when ApiKeyOptions.RunMigrationsOnStartup.
        services.AddHostedService<ApiKeyMigrationHostedService>();
@@ -1,5 +1,6 @@
 using System.Globalization;
 using Microsoft.Data.Sqlite;
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;

 namespace ZB.MOM.WW.Auth.ApiKeys.Sqlite;

@@ -11,7 +12,7 @@ public sealed class AuthStoreMigrationException(string message) : InvalidOperati
 /// is safe to run repeatedly. Refuses to run against a database whose on-disk version
 /// is newer than this build supports.
 /// </summary>
-public sealed class SqliteAuthStoreMigrator(AuthSqliteConnectionFactory connectionFactory)
+public sealed class SqliteAuthStoreMigrator(AuthSqliteConnectionFactory connectionFactory) : IApiKeyStoreMigrator
 {
    /// <summary>Applies the schema migration to the auth store.</summary>
    /// <param name="cancellationToken">Cancellation token.</param>
@@ -38,6 +38,9 @@

  <ItemGroup>
    <InternalsVisibleTo Include="ZB.MOM.WW.Auth.ApiKeys.Tests" />
+    <!-- Task 5's SqliteApiKeyStoreImporter parses SQLite values with SqliteValueParsing (internal)
+         rather than duplicating the parser. -->
+    <InternalsVisibleTo Include="ZB.MOM.WW.Auth.ApiKeys.SqlServer" />
  </ItemGroup>

 </Project>
@@ -49,11 +49,16 @@ public static class ServiceCollectionExtensions
        // Fail fast at startup on a misconfigured directory rather than on first login.
        services.AddSingleton<IValidateOptions<LdapOptions>, LdapOptionsValidator>();

-        // LdapAuthService is stateless: it holds only a snapshot of LdapOptions and a stateless
-        // connection factory, and opens/disposes a connection per call. It is not IDisposable.
-        // Singleton is correct; TryAdd mirrors the pattern in AddZbApiKeyAuth (idempotency).
+        // LdapAuthService opens/disposes a connection per call and is not IDisposable. Its only
+        // mutable state is the sticky endpoint preference used by FallbackServers failover —
+        // which is precisely why the instance must be shared, so a primary-DC outage is paid
+        // for once rather than on every login. Singleton is correct; TryAdd mirrors the pattern
+        // in AddZbApiKeyAuth (idempotency). The logger is resolved with GetService, not
+        // GetRequiredService, so a host without logging configured still resolves the service.
        services.TryAddSingleton<ILdapAuthService>(sp =>
-            new LdapAuthService(sp.GetRequiredService<IOptions<LdapOptions>>().Value));
+            new LdapAuthService(
+                sp.GetRequiredService<IOptions<LdapOptions>>().Value,
+                sp.GetService<Microsoft.Extensions.Logging.ILogger<LdapAuthService>>()));

        return services;
    }
@@ -0,0 +1,60 @@
+namespace ZB.MOM.WW.Auth.Ldap.Internal;
+
+/// <summary>A single LDAP server endpoint resolved from configuration.</summary>
+internal sealed record LdapEndpoint(string Host, int Port)
+{
+    /// <summary>Renders the endpoint as <c>host:port</c> — the form used in failover log messages.</summary>
+    public override string ToString() => $"{Host}:{Port}";
+}
+
+/// <summary>
+/// Parses <c>FallbackServers</c> entries: <c>"host"</c> (adopts the caller's default port)
+/// or <c>"host:port"</c>. IPv6 literals are not supported — use a DNS name.
+/// </summary>
+internal static class LdapEndpointParser
+{
+    /// <summary>
+    /// Attempts to parse a single <c>FallbackServers</c> entry.
+    /// </summary>
+    /// <param name="entry">The configured entry, <c>"host"</c> or <c>"host:port"</c>. Surrounding whitespace is tolerated.</param>
+    /// <param name="defaultPort">Port adopted when the entry carries no explicit one.</param>
+    /// <param name="endpoint">The parsed endpoint on success; undefined otherwise.</param>
+    /// <param name="error">A human-readable reason on failure; empty on success.</param>
+    /// <returns><see langword="true"/> when the entry parsed.</returns>
+    public static bool TryParse(string entry, int defaultPort, out LdapEndpoint endpoint, out string error)
+    {
+        endpoint = null!;
+        error = "";
+
+        if (string.IsNullOrWhiteSpace(entry))
+        {
+            error = "entry is empty or whitespace";
+            return false;
+        }
+
+        var parts = entry.Trim().Split(':');
+        if (parts.Length > 2)
+        {
+            error = $"entry \"{entry.Trim()}\" contains more than one ':' " +
+                    "(IPv6 literals are not supported — use a DNS name)";
+            return false;
+        }
+
+        var host = parts[0].Trim();
+        if (host.Length == 0)
+        {
+            error = $"entry \"{entry.Trim()}\" has an empty host";
+            return false;
+        }
+
+        var port = defaultPort;
+        if (parts.Length == 2 && (!int.TryParse(parts[1].Trim(), out port) || port is < 1 or > 65535))
+        {
+            error = $"entry \"{entry.Trim()}\" has an invalid port (must be 1–65535)";
+            return false;
+        }
+
+        endpoint = new LdapEndpoint(host, port);
+        return true;
+    }
+}
@@ -0,0 +1,60 @@
+namespace ZB.MOM.WW.Auth.Ldap.Internal;
+
+/// <summary>
+/// Reduces a typed login to the bare account name used by <see cref="Abstractions.Ldap.LdapBindMode.User"/>:
+/// the qualifier is stripped before the name reaches the bind template and the search filter, so audit
+/// rows never carry <c>DOMAIN\jdoe</c> for one sign-in and <c>jdoe</c> for the next. Case is NOT
+/// normalised here: the returned username is canonicalised from the directory entry's
+/// <c>UserNameAttribute</c> value, so <c>JDoe</c> and <c>jdoe</c> also yield one identity; this name is
+/// only the fallback when the entry has no such value.
+/// </summary>
+internal static class LdapUserName
+{
+    /// <summary>
+    /// Trims <paramref name="input"/> and, when <paramref name="acceptDomainQualified"/>, strips one
+    /// leading <c>DOMAIN\</c> or one trailing <c>@suffix</c>. Returns <see langword="false"/> (with
+    /// <paramref name="name"/> empty) for blank input, for both forms at once, for a repeated
+    /// separator, for an empty side, or for any qualified form when <paramref name="acceptDomainQualified"/>
+    /// is <see langword="false"/>.
+    /// </summary>
+    public static bool TryNormalize(string? input, bool acceptDomainQualified, out string name)
+    {
+        name = "";
+        if (string.IsNullOrWhiteSpace(input))
+            return false;
+
+        var s = input.Trim();
+        var slash = s.IndexOf('\\');
+        var at = s.IndexOf('@');
+
+        if (slash >= 0 || at >= 0)
+        {
+            if (!acceptDomainQualified || (slash >= 0 && at >= 0))
+                return false;
+
+            if (slash >= 0)
+            {
+                if (slash == 0 || s.IndexOf('\\', slash + 1) >= 0)
+                    return false;
+                s = s[(slash + 1)..];
+            }
+            else
+            {
+                // Unlike the slash branch (whose extracted suffix is checked for emptiness by the
+                // final length check below), the extracted piece here is the *prefix* — so an empty
+                // domain part (a trailing "@", e.g. "jdoe@") must be rejected explicitly.
+                if (at == s.Length - 1 || s.IndexOf('@', at + 1) >= 0)
+                    return false;
+                s = s[..at];
+            }
+
+            s = s.Trim();
+        }
+
+        if (s.Length == 0)
+            return false;
+
+        name = s;
+        return true;
+    }
+}
@@ -99,23 +99,73 @@ internal sealed class NovellLdapConnection : ILdapConnection
            attributes.ToArray(),
            typesOnly: false);

+        return ReadEntries(results);
+    }
+
+    /// <summary>
+    /// Drains a result set into entries, <b>skipping continuation referrals</b>.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// Active Directory answers any search based at a domain root with one or more
+    /// continuation references (to <c>CN=Configuration</c>, <c>DC=ForestDnsZones</c> and
+    /// friends). With referral following disabled — the default, and what we want, since
+    /// chasing a referral would re-bind somewhere we did not authenticate — Novell surfaces
+    /// each one by throwing <see cref="LdapReferralException"/> from <see cref="ILdapSearchResults.Next"/>.
+    /// </para>
+    /// <para>
+    /// A continuation reference is a <em>pointer to another naming context</em>, not an error:
+    /// it says "there may be more results over there", never that this search failed. Letting
+    /// it escape made the whole search stage fail, which <c>LdapAuthService</c> maps to
+    /// <see cref="Abstractions.Ldap.LdapAuthFailure.ServiceAccountBindFailed"/> — so against a
+    /// real AD domain root, every login failed with a system-side error while the directory,
+    /// the credentials and both DCs were perfectly healthy. Verified against
+    /// <c>DC=nam,DC=zimmer,DC=com</c> on 2026-08-12.
+    /// </para>
+    /// <para>
+    /// Skipping is the correct behaviour rather than a workaround: the entries we can see are
+    /// exactly the ones in the naming context we bound to, which is the set the caller asked
+    /// about. Internal so the referral handling can be unit-tested through the
+    /// <see cref="ILdapSearchResults"/> seam without a live directory.
+    /// </para>
+    /// </remarks>
+    internal static List<LdapSearchEntry> ReadEntries(ILdapSearchResults results)
+    {
        var entries = new List<LdapSearchEntry>();
-        while (results.HasMore())
+        var referrals = 0;
+
+        while (true)
        {
-            var entry = results.Next();
-            var attrs = new Dictionary<string, IReadOnlyList<string>>(StringComparer.OrdinalIgnoreCase);
-
-            foreach (LdapAttribute attr in entry.GetAttributeSet())
+            try
            {
-                attrs[attr.Name] = attr.StringValueArray.ToList();
-            }
+                if (!results.HasMore()) break;

-            entries.Add(new LdapSearchEntry(entry.Dn, attrs));
+                var entry = results.Next();
+                var attrs = new Dictionary<string, IReadOnlyList<string>>(StringComparer.OrdinalIgnoreCase);
+
+                foreach (LdapAttribute attr in entry.GetAttributeSet())
+                {
+                    attrs[attr.Name] = attr.StringValueArray.ToList();
+                }
+
+                entries.Add(new LdapSearchEntry(entry.Dn, attrs));
+            }
+            catch (LdapException ex) when (ex is LdapReferralException || ex.ResultCode == LdapException.Referral)
+            {
+                // Novell dequeues the reference before throwing, so the enumeration advances and
+                // this terminates. The cap is a belt-and-braces guard: a client that threw without
+                // advancing would otherwise spin forever inside a fail-closed auth path, which is a
+                // far worse failure than under-reporting an implausible number of referrals.
+                if (++referrals > MaxReferrals) break;
+            }
        }

        return entries;
    }

+    /// <summary>Upper bound on continuation references tolerated in one search — see <see cref="ReadEntries"/>.</summary>
+    private const int MaxReferrals = 1_000;
+
    /// <inheritdoc/>
    public void Dispose()
    {
@@ -1,3 +1,4 @@
+using Microsoft.Extensions.Logging;
 using Novell.Directory.Ldap;
 using ZB.MOM.WW.Auth.Abstractions.Ldap;
 using ZB.MOM.WW.Auth.Ldap.Internal;
@@ -11,6 +12,7 @@ namespace ZB.MOM.WW.Auth.Ldap;
 /// <see cref="ILdapConnection"/> so the logic is unit-testable without a live server.
 /// </summary>
 /// <remarks>
+/// <para>
 /// Fully fail-closed: authentication never throws to the caller — every error, expected
 /// or unexpected, is mapped to a structured <see cref="LdapAuthResult.Fail(LdapAuthFailure)"/>.
 /// A success is only returned for a user that resolved to exactly one entry, whose password
@@ -18,11 +20,42 @@ namespace ZB.MOM.WW.Auth.Ldap;
 /// Service-account bind failures (<see cref="LdapAuthFailure.ServiceAccountBindFailed"/>) are
 /// kept distinct from end-user bind failures (<see cref="LdapAuthFailure.BadCredentials"/>) so
 /// a system misconfiguration is not mistaken for bad user input.
+/// </para>
+/// <para>
+/// When <see cref="LdapOptions.FallbackServers"/> is non-empty the flow above runs against
+/// an ordered endpoint list (primary first, then fallbacks in declared order) and moves on
+/// to the next endpoint ONLY on a system-side failure — connect/TLS, service-account bind,
+/// or search — i.e. when the endpoint gave no authoritative answer. Any answer from a healthy
+/// directory (success, or a user-side failure such as
+/// <see cref="LdapAuthFailure.BadCredentials"/> / <see cref="LdapAuthFailure.UserNotFound"/>)
+/// is final and is never retried elsewhere: retrying a rejected password against a second DC
+/// would turn one bad-password event into several, tripping account lockouts. Endpoint
+/// preference is sticky — the last endpoint that answered is tried first next time — so a
+/// primary outage costs its connect timeout once, not once per login.
+/// </para>
+/// <para>
+/// The service is safe to share (it is registered as a singleton). The sticky index is the
+/// only mutable state; concurrent logins may race on it, which is benign — the worst outcome
+/// is one login starting at a stale endpoint and walking on.
+/// </para>
+/// <para>
+/// With <c>BindMode = User</c> the service-account bind and the user re-bind are replaced by a
+/// single bind as the user (<see cref="LdapOptions.UserBindTemplate"/>); see <see cref="CompleteAsUser"/>.
+/// </para>
 /// </remarks>
 public sealed class LdapAuthService : ILdapAuthService
 {
    private readonly LdapOptions _options;
    private readonly ILdapConnectionFactory _connectionFactory;
+    private readonly IReadOnlyList<LdapEndpoint> _endpoints;
+    private readonly ILogger<LdapAuthService>? _logger;
+
+    /// <summary>
+    /// Index into <see cref="_endpoints"/> of the endpoint tried first. Written on any
+    /// authoritative answer from a non-preferred endpoint. Accessed via
+    /// <see cref="Volatile"/> only; benign races are accepted (see the type remarks).
+    /// </summary>
+    private int _preferredIndex;

    /// <summary>
    /// Production constructor: binds against a live directory via the real
@@ -31,9 +64,15 @@ public sealed class LdapAuthService : ILdapAuthService
    /// <see cref="LdapOptions.AllowInsecure"/> bypass) is carried into the factory so each
    /// connection is built with it.
    /// </summary>
-    public LdapAuthService(LdapOptions options)
+    /// <param name="options">The directory connection and search settings.</param>
+    /// <param name="logger">
+    /// Optional. When supplied, endpoint failover is made visible: a Warning per failed
+    /// endpoint hop and an Information line when the sticky preference moves. Omitting it
+    /// (the default) leaves the service silent — hosts without logging still work.
+    /// </param>
+    public LdapAuthService(LdapOptions options, ILogger<LdapAuthService>? logger = null)
        : this(options, new NovellLdapConnectionFactory(
-            options.AllowInsecure, options.ServerCertificateValidationCallback))
+            options.AllowInsecure, options.ServerCertificateValidationCallback), logger)
    {
    }

@@ -42,10 +81,33 @@ public sealed class LdapAuthService : ILdapAuthService
    /// so the bind/search logic can be exercised without a live directory. Internal
    /// because the connection seam is an implementation detail.
    /// </summary>
-    internal LdapAuthService(LdapOptions options, ILdapConnectionFactory connectionFactory)
+    internal LdapAuthService(
+        LdapOptions options,
+        ILdapConnectionFactory connectionFactory,
+        ILogger<LdapAuthService>? logger = null)
    {
        _options = options;
        _connectionFactory = connectionFactory;
+        _logger = logger;
+        _endpoints = BuildEndpoints(options);
+    }
+
+    /// <summary>
+    /// Primary first, then fallbacks in declared order. The validator rejects malformed
+    /// entries at boot; unparseable entries are skipped defensively here so a service
+    /// constructed without options validation degrades to the parseable subset rather
+    /// than throwing from a constructor.
+    /// </summary>
+    private static IReadOnlyList<LdapEndpoint> BuildEndpoints(LdapOptions options)
+    {
+        var endpoints = new List<LdapEndpoint> { new(options.Server, options.Port) };
+        foreach (var entry in options.FallbackServers)
+        {
+            if (LdapEndpointParser.TryParse(entry, options.Port, out var endpoint, out _))
+                endpoints.Add(endpoint);
+        }
+
+        return endpoints;
    }

    /// <inheritdoc />
@@ -54,7 +116,10 @@ public sealed class LdapAuthService : ILdapAuthService
    /// bind-then-search-then-bind flow is wrapped so that any error — expected or unexpected —
    /// is mapped to a structured <see cref="LdapAuthResult.Fail(LdapAuthFailure)"/>. A
    /// <c>Succeeded == true</c> result is only ever returned when the user resolved to exactly
-    /// one entry, their password verified, AND at least one group was extracted.
+    /// one entry, their password verified, AND at least one group was extracted. With
+    /// <see cref="LdapOptions.FallbackServers"/> configured the flow may run against more than
+    /// one endpoint before answering — see the type-level remarks for the walk and its
+    /// system-side-only failover rule.
    /// </remarks>
    public Task<LdapAuthResult> AuthenticateAsync(string username, string password, CancellationToken ct)
    {
@@ -80,25 +145,105 @@ public sealed class LdapAuthService : ILdapAuthService
        //    filter and the returned result (avoids two identities for one person).
        username = username.Trim();

+        // 3b. An empty password must never reach the directory, in EITHER bind mode: AD answers a
+        //     simple bind with an empty password as an UNAUTHENTICATED bind and reports success —
+        //     the user bind (or the ServiceAccount-mode re-bind as the user's DN) would admit anyone.
+        //     Refused here, before any network call, so no consumer depends on its own edge check.
+        if (string.IsNullOrEmpty(password))
+            return LdapAuthResult.Fail(LdapAuthFailure.BadCredentials);
+
+        // 3c. User-bind mode: reduce the name to the bare account name the bind template and the
+        //     search filter share. (The returned username is canonicalised from the entry later.)
+        if (_options.BindMode == LdapBindMode.User
+            && !LdapUserName.TryNormalize(username, _options.AcceptDomainQualifiedNames, out username))
+        {
+            return LdapAuthResult.Fail(LdapAuthFailure.BadCredentials);
+        }
+
+        // 4. Walk the endpoint list, preferred one first. A single-endpoint configuration
+        //    (no FallbackServers) collapses to exactly one attempt — unchanged behaviour.
+        var preferred = Volatile.Read(ref _preferredIndex);
+        if ((uint)preferred >= (uint)_endpoints.Count)
+            preferred = 0; // defensive: the list is immutable, but never index out of it
+
+        // Seeded so an exhausted walk still returns the conservative system-side bucket
+        // even in the impossible case of an empty endpoint list.
+        var lastSystemSideFailure = LdapAuthResult.Fail(LdapAuthFailure.ServiceAccountBindFailed);
+
+        foreach (var index in WalkOrder(_endpoints.Count, preferred))
+        {
+            var result = AuthenticateAgainst(_endpoints[index], username, password);
+
+            // ServiceAccountBindFailed is the (sole) system-side bucket: this endpoint gave
+            // no authoritative answer, so try the next one. Every other outcome — success, or
+            // a user-side failure like UserNotFound/BadCredentials — came from a healthy
+            // directory and is FINAL: re-asking a second DC about a rejected password would
+            // multiply one bad-password event into several against the account lockout policy.
+            if (result.Failure == LdapAuthFailure.ServiceAccountBindFailed)
+            {
+                lastSystemSideFailure = result;
+                continue;
+            }
+
+            // Sticky preference: any authoritative answer proves this endpoint healthy.
+            // Updating only on Succeeded would re-pay the dead primary's connect timeout on
+            // every mistyped password for the whole outage.
+            if (index != preferred)
+            {
+                Volatile.Write(ref _preferredIndex, index);
+                _logger?.LogInformation("LDAP sticky preference moved to {Endpoint}.", _endpoints[index]);
+            }
+
+            return result;
+        }
+
+        // Every endpoint failed system-side. The preference is deliberately left where it was:
+        // a total outage says nothing about which endpoint to prefer once one recovers.
+        return lastSystemSideFailure;
+    }
+
+    /// <summary>Preferred endpoint first, then every other endpoint in declared order.</summary>
+    private static IEnumerable<int> WalkOrder(int count, int preferred)
+    {
+        yield return preferred;
+        for (var i = 0; i < count; i++)
+        {
+            if (i != preferred)
+                yield return i;
+        }
+    }
+
+    /// <summary>
+    /// Runs one full bind-then-search-then-bind attempt against a single endpoint on a fresh
+    /// connection. The fail-closed contract holds per attempt: this never throws, and it
+    /// returns <see cref="LdapAuthFailure.ServiceAccountBindFailed"/> for — and only for —
+    /// system-side faults, which is what the caller reads as "no authoritative answer, try
+    /// the next endpoint".
+    /// </summary>
+    private LdapAuthResult AuthenticateAgainst(LdapEndpoint endpoint, string username, string password)
+    {
        // The whole flow runs inside an outer fail-closed guard: a StageFailure carries an
        // already-mapped failure out of a stage, and any OTHER unexpected exception defaults to
        // the most conservative system-side bucket. Either way the caller gets a structured result.
        try
        {
+            // A fresh connection per attempt: a Novell connection that has failed to connect
+            // or bind is not safely reusable against another host.
            using var conn = _connectionFactory.Create();

-            // 4. Open the connection (transport/TLS handling lives in the adapter). The
+            // 5. Open the connection (transport/TLS handling lives in the adapter). The
            //    per-operation timeout (ConnectionTimeoutMs) is applied by the adapter here.
-            //    A failure to connect/upgrade means the directory is unreachable — a
-            //    system-side fault, not the user's, so map it to ServiceAccountBindFailed.
+            //    A failure to connect/upgrade means this endpoint is unreachable — a
+            //    system-side fault, not the user's, so map it to ServiceAccountBindFailed
+            //    and let the caller try the next endpoint.
            // NOTE: the LdapAuthFailure enum has no dedicated DirectoryUnavailable value;
            //       ServiceAccountBindFailed is the closest system-side bucket. A future
            //       Abstractions change could add DirectoryUnavailable to disambiguate.
            try
            {
                conn.Connect(
-                    _options.Server,
-                    _options.Port,
+                    endpoint.Host,
+                    endpoint.Port,
                    _options.Transport,
                    _options.AllowInsecure,
                    _options.ConnectionTimeoutMs,
@@ -106,110 +251,197 @@ public sealed class LdapAuthService : ILdapAuthService
            }
            catch (LdapException)
            {
-                throw new StageFailure(LdapAuthFailure.ServiceAccountBindFailed);
+                throw new StageFailure(LdapAuthFailure.ServiceAccountBindFailed, "connect");
            }

-            // 5. Service-account bind so we can search for the user's DN. A bind failure
-            //    here is a service-account misconfiguration — DISTINCT from a user-credential
-            //    failure — so it maps to ServiceAccountBindFailed.
-            try
-            {
-                conn.Bind(_options.ServiceAccountDn, _options.ServiceAccountPassword);
-            }
-            catch (LdapException)
-            {
-                throw new StageFailure(LdapAuthFailure.ServiceAccountBindFailed);
-            }
-
-            // 6. Search for the user entry by the configured username attribute. A search
-            //    failure is infrastructure (directory unreachable / unhealthy) — system-side.
-            // NOTE: same enum limitation as Connect above; ServiceAccountBindFailed is the
-            //       closest system-side bucket until a DirectoryUnavailable value exists.
-            IReadOnlyList<LdapSearchEntry> entries;
-            try
-            {
-                var filter = $"({_options.UserNameAttribute}={LdapEscaping.Filter(username)})";
-                entries = conn.Search(_options.SearchBase, filter, BuildSearchAttributes());
-            }
-            catch (LdapException)
-            {
-                throw new StageFailure(LdapAuthFailure.ServiceAccountBindFailed);
-            }
-
-            // 7. Require exactly one match. Zero -> UserNotFound; two or more -> AmbiguousUser.
-            //    We never attempt a user bind against an ambiguous DN.
-            if (entries.Count == 0)
-                return LdapAuthResult.Fail(LdapAuthFailure.UserNotFound);
-            if (entries.Count >= 2)
-                return LdapAuthResult.Fail(LdapAuthFailure.AmbiguousUser);
-
-            var entry = entries[0];
-
-            // 8. User bind: re-bind as the resolved DN to verify the password. A bind failure
-            //    here is the end user's bad credentials.
-            try
-            {
-                conn.Bind(entry.Dn, password);
-            }
-            catch (LdapException)
-            {
-                throw new StageFailure(LdapAuthFailure.BadCredentials);
-            }
-
-            // 9. Group extraction. Fail closed: an empty/missing group set, or any error while
-            //    extracting groups, is a GroupLookupFailed — never a zero-group success.
-            IReadOnlyList<string> groups;
-            try
-            {
-                groups = ExtractGroups(entry);
-            }
-            catch
-            {
-                throw new StageFailure(LdapAuthFailure.GroupLookupFailed);
-            }
-
-            if (groups.Count == 0)
-                return LdapAuthResult.Fail(LdapAuthFailure.GroupLookupFailed);
-
-            // 10. Success — and only here, with a verified password and >= 1 group.
-            var displayName = ExtractDisplayName(entry, username);
-            return LdapAuthResult.Success(username, displayName, groups);
+            return _options.BindMode == LdapBindMode.User
+                ? CompleteAsUser(conn, username, password)
+                : CompleteWithServiceAccount(conn, username, password);
        }
        catch (StageFailure stage)
        {
-            // A stage mapped its own failure; surface it as a structured result.
+            // A stage mapped its own failure; surface it as a structured result. Log at Warning
+            // so a silent failover — the failure mode this feature exists to survive — is
+            // visible in the host's log rather than only in a latency graph.
+            _logger?.LogWarning(
+                "LDAP endpoint {Endpoint} failed system-side at {Stage}; treating as no answer.",
+                endpoint, stage.Stage);
            return LdapAuthResult.Fail(stage.Failure);
        }
-        catch
+        catch (Exception ex)
        {
            // Belt-and-braces: ANY unexpected exception fails closed to the most conservative
            // system-side bucket rather than propagating to the caller.
+            _logger?.LogWarning(ex, "LDAP endpoint {Endpoint} failed system-side with an unexpected error.", endpoint);
            return LdapAuthResult.Fail(LdapAuthFailure.ServiceAccountBindFailed);
        }
    }

+    /// <summary>Steps 6–11 of the 0.3.0 flow: service-account bind, search, user re-bind, groups.</summary>
+    private LdapAuthResult CompleteWithServiceAccount(ZB.MOM.WW.Auth.Ldap.Internal.ILdapConnection conn, string username, string password)
+    {
+        // 6. Service-account bind so we can search for the user's DN. A bind failure here is a
+        //    service-account misconfiguration — DISTINCT from a user-credential failure — so it
+        //    maps to ServiceAccountBindFailed.
+        try
+        {
+            conn.Bind(_options.ServiceAccountDn, _options.ServiceAccountPassword);
+        }
+        catch (LdapException)
+        {
+            throw new StageFailure(LdapAuthFailure.ServiceAccountBindFailed, "service-bind");
+        }
+
+        // 7–8. Require exactly one match. Zero -> UserNotFound; two or more -> AmbiguousUser.
+        //      Both are authoritative answers from a healthy directory, so they end the walk.
+        //      We never attempt a user bind against an ambiguous DN.
+        var entries = SearchUser(conn, username);
+        if (entries.Count == 0)
+            return LdapAuthResult.Fail(LdapAuthFailure.UserNotFound);
+        if (entries.Count >= 2)
+            return LdapAuthResult.Fail(LdapAuthFailure.AmbiguousUser);
+
+        var entry = entries[0];
+
+        // 9. User bind: re-bind as the resolved DN to verify the password. A bind failure here is
+        //    the end user's bad credentials — the directory answered, so this is returned directly
+        //    and never retried against another endpoint.
+        try
+        {
+            conn.Bind(entry.Dn, password);
+        }
+        catch (LdapException)
+        {
+            return LdapAuthResult.Fail(LdapAuthFailure.BadCredentials);
+        }
+
+        return SucceedFromEntry(entry, username);
+    }
+
+    /// <summary>
+    /// <see cref="LdapBindMode.User"/>: the user's own bind is the authentication; the same
+    /// connection then reads their entry. Only result 49 (invalid credentials — AD's answer for a
+    /// wrong password AND for unknown, locked, disabled or expired accounts) is the user's failure
+    /// and final; any other bind error means this endpoint gave no answer, so the walk moves on.
+    /// </summary>
+    private LdapAuthResult CompleteAsUser(ZB.MOM.WW.Auth.Ldap.Internal.ILdapConnection conn, string username, string password)
+    {
+        var bindName = _options.UserBindTemplate.Replace("{0}", username, StringComparison.Ordinal);
+        try
+        {
+            conn.Bind(bindName, password);
+        }
+        catch (LdapException ex) when (ex.ResultCode == LdapException.InvalidCredentials)
+        {
+            return LdapAuthResult.Fail(LdapAuthFailure.BadCredentials);
+        }
+        catch (LdapException)
+        {
+            throw new StageFailure(LdapAuthFailure.ServiceAccountBindFailed, "user-bind");
+        }
+
+        // After a good bind, anything but exactly one readable entry is a failure to establish the
+        // user's groups — never a success, and never UserNotFound (the directory just accepted them).
+        var entries = SearchUser(conn, username);
+        if (entries.Count != 1)
+            return LdapAuthResult.Fail(LdapAuthFailure.GroupLookupFailed);
+
+        return SucceedFromEntry(entries[0], username);
+    }
+
+    /// <summary>Step 7: search for the user entry; a search fault is system-side.</summary>
+    private IReadOnlyList<LdapSearchEntry> SearchUser(ZB.MOM.WW.Auth.Ldap.Internal.ILdapConnection conn, string username)
+    {
+        // A search failure is infrastructure (directory unreachable / unhealthy) — system-side.
+        // NOTE: the LdapAuthFailure enum has no DirectoryUnavailable value, so a failed search maps
+        //       to ServiceAccountBindFailed, the closest system-side bucket (same limitation as
+        //       Connect), until a DirectoryUnavailable value exists.
+        try
+        {
+            var filter = $"({_options.UserNameAttribute}={LdapEscaping.Filter(username)})";
+            return conn.Search(_options.SearchBase, filter, BuildSearchAttributes());
+        }
+        catch (LdapException)
+        {
+            throw new StageFailure(LdapAuthFailure.ServiceAccountBindFailed, "search");
+        }
+    }
+
+    /// <summary>
+    /// Steps 10–11: fail-closed group extraction, then success. The returned username is the
+    /// entry's own <see cref="LdapOptions.UserNameAttribute"/> value (the directory's spelling), so
+    /// <c>JDoe</c> and <c>jdoe</c> sign in as one identity; the typed (normalised) name is the
+    /// fallback when the entry carries no such value.
+    /// </summary>
+    private LdapAuthResult SucceedFromEntry(LdapSearchEntry entry, string username)
+    {
+        IReadOnlyList<string> groups;
+        try
+        {
+            groups = ExtractGroups(entry);
+        }
+        catch
+        {
+            return LdapAuthResult.Fail(LdapAuthFailure.GroupLookupFailed);
+        }
+
+        if (groups.Count == 0)
+            return LdapAuthResult.Fail(LdapAuthFailure.GroupLookupFailed);
+
+        var canonical = ExtractUserName(entry, username);
+        return LdapAuthResult.Success(canonical, ExtractDisplayName(entry, canonical), groups);
+    }
+
    /// <summary>
    /// Internal control-flow exception that carries an already-mapped <see cref="LdapAuthFailure"/>
    /// out of a stage to the single fail-closed catch site. Never escapes this type.
    /// </summary>
    private sealed class StageFailure : Exception
    {
-        public StageFailure(LdapAuthFailure failure) => Failure = failure;
+        public StageFailure(LdapAuthFailure failure, string stage)
+        {
+            Failure = failure;
+            Stage = stage;
+        }

        public LdapAuthFailure Failure { get; }
+
+        /// <summary>Which stage faulted (<c>connect</c> / <c>service-bind</c> / <c>user-bind</c> / <c>search</c>), for diagnostics.</summary>
+        public string Stage { get; }
    }

    /// <summary>
-    /// Builds the distinct attribute list requested from the directory. The display-name
-    /// and group attributes are de-duplicated so we never request the same attribute twice
-    /// when an operator configures them to the same value.
+    /// Builds the distinct attribute list requested from the directory: display name, groups and
+    /// the username attribute (read back as the canonical username). They are de-duplicated,
+    /// case-insensitively, so we never request the same attribute twice when an operator
+    /// configures two of them to the same value.
    /// </summary>
    private IReadOnlyList<string> BuildSearchAttributes()
    {
-        if (string.Equals(_options.DisplayNameAttribute, _options.GroupAttribute, StringComparison.OrdinalIgnoreCase))
-            return new[] { _options.DisplayNameAttribute };
+        var attributes = new List<string>(3);
+        foreach (var name in new[] { _options.DisplayNameAttribute, _options.GroupAttribute, _options.UserNameAttribute })
+        {
+            if (!attributes.Contains(name, StringComparer.OrdinalIgnoreCase))
+                attributes.Add(name);
+        }

-        return new[] { _options.DisplayNameAttribute, _options.GroupAttribute };
+        return attributes;
+    }
+
+    /// <summary>
+    /// Returns the first non-blank value of the configured username attribute — the directory's
+    /// canonical spelling — falling back to the (already normalised) typed username.
+    /// </summary>
+    private string ExtractUserName(LdapSearchEntry entry, string username)
+    {
+        if (entry.Attributes.TryGetValue(_options.UserNameAttribute, out var values)
+            && values.Count > 0
+            && !string.IsNullOrWhiteSpace(values[0]))
+        {
+            return values[0];
+        }
+
+        return username;
    }

    /// <summary>
@@ -10,7 +10,7 @@ namespace ZB.MOM.WW.Auth.Ldap;
 /// </summary>
 /// <remarks>
 /// Validation is skipped entirely when <see cref="LdapOptions.Enabled"/> is <c>false</c>
-/// (a disabled provider's connection fields are inert). When enabled, four conditions
+/// (a disabled provider's connection fields are inert). When enabled, five conditions
 /// are enforced:
 /// <list type="bullet">
 /// <item>plaintext transport (<see cref="LdapTransport.None"/>) is rejected unless
@@ -18,8 +18,13 @@ namespace ZB.MOM.WW.Auth.Ldap;
 /// <item><see cref="LdapOptions.Server"/> must be specified (no sane default host);</item>
 /// <item><see cref="LdapOptions.SearchBase"/> must be specified (the DN root every
 /// search runs against);</item>
-/// <item><see cref="LdapOptions.ServiceAccountDn"/> must be specified — an empty value
-/// would bind anonymously, defeating the search-then-bind authentication flow.</item>
+/// <item>in <see cref="LdapBindMode.ServiceAccount"/> mode <see cref="LdapOptions.ServiceAccountDn"/>
+/// must be specified — an empty value would bind anonymously, defeating the search-then-bind
+/// authentication flow; in <see cref="LdapBindMode.User"/> mode <see cref="LdapOptions.UserBindTemplate"/>
+/// must contain <c>{0}</c> exactly once instead;</item>
+/// <item>every <see cref="LdapOptions.FallbackServers"/> entry must parse as
+/// <c>"host"</c> or <c>"host:port"</c> — a typo'd backup DC must not lie dormant
+/// until the primary outage it exists to survive.</item>
 /// </list>
 /// </remarks>
 public sealed class LdapOptionsValidator : IValidateOptions<LdapOptions>
@@ -60,7 +65,20 @@ public sealed class LdapOptionsValidator : IValidateOptions<LdapOptions>
                "set it to the search-base DN (e.g. \"dc=example,dc=com\").");
        }

-        if (string.IsNullOrWhiteSpace(options.ServiceAccountDn))
+        if (options.BindMode == LdapBindMode.User)
+        {
+            // User mode binds as the signing-in user, so there is no service account to require.
+            // The template is the bind identity: without exactly one {0} every login would bind as
+            // the same fixed name (or as a name with the user spliced in twice).
+            if (CountPlaceholders(options.UserBindTemplate) != 1)
+            {
+                return ValidateOptionsResult.Fail(
+                    $"{nameof(LdapOptions.UserBindTemplate)} must contain {{0}} exactly once when " +
+                    $"{nameof(LdapOptions.BindMode)} is {nameof(LdapBindMode.User)} — e.g. \"{{0}}@example.com\" " +
+                    "or \"EXAMPLE\\{0}\".");
+            }
+        }
+        else if (string.IsNullOrWhiteSpace(options.ServiceAccountDn))
        {
            return ValidateOptionsResult.Fail(
                $"{nameof(LdapOptions.ServiceAccountDn)} is required but was empty or whitespace — " +
@@ -68,6 +86,33 @@ public sealed class LdapOptionsValidator : IValidateOptions<LdapOptions>
                "(e.g. \"cn=svc,dc=example,dc=com\").");
        }

+        // Fallback endpoints are walked by LdapAuthService on system-side failures. A
+        // malformed entry would be silently skipped there (the walk degrades to the
+        // parseable subset rather than throwing), so it must be caught at boot instead —
+        // otherwise an operator's typo'd backup DC would only surface as a failed
+        // failover during the primary outage it was meant to survive.
+        foreach (var entry in options.FallbackServers)
+        {
+            if (!Internal.LdapEndpointParser.TryParse(entry, options.Port, out _, out var error))
+            {
+                return ValidateOptionsResult.Fail(
+                    $"{nameof(LdapOptions.FallbackServers)} contains an invalid entry — {error}. " +
+                    "Each entry must be \"host\" or \"host:port\" (port 1–65535).");
+            }
+        }
+
        return ValidateOptionsResult.Success;
    }
+
+    private static int CountPlaceholders(string? template)
+    {
+        if (string.IsNullOrEmpty(template))
+            return 0;
+
+        var count = 0;
+        for (var i = template.IndexOf("{0}", StringComparison.Ordinal); i >= 0;
+             i = template.IndexOf("{0}", i + 3, StringComparison.Ordinal))
+            count++;
+        return count;
+    }
 }
@@ -22,6 +22,7 @@
  <ItemGroup>
    <PackageReference Include="Novell.Directory.Ldap.NETStandard" />
    <PackageReference Include="Microsoft.Extensions.Options" />
+    <PackageReference Include="Microsoft.Extensions.Logging.Abstractions" />
  </ItemGroup>

  <ItemGroup>
@@ -0,0 +1,101 @@
+using Microsoft.Data.SqlClient;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests.Live;
+
+/// <summary>
+/// Shared connection details for the env-gated live SQL Server suite.
+/// </summary>
+/// <remarks>
+/// Gated on <c>AUTH_SQLSERVER_CONNSTR</c> and skips cleanly when it is unset, matching the family
+/// live-test idiom (see <c>ZB.MOM.WW.Secrets.Replicator.SqlServer</c>) — the offline suite must stay
+/// runnable on any machine. Set it to a database the test is allowed to create and drop a schema in,
+/// for example <c>Server=localhost,1433;Database=ZbAuthTest;User Id=sa;Password=...;TrustServerCertificate=true</c>.
+/// Each test class works in a throwaway <em>schema</em> rather than a throwaway database: it needs no
+/// CREATE DATABASE right, and a virgin schema still reproduces the CREATE SCHEMA race the migrator's
+/// retry loop exists for.
+/// </remarks>
+public static class LiveSqlServer
+{
+    /// <summary>Environment variable holding the live connection string.</summary>
+    public const string ConnectionStringVariable = "AUTH_SQLSERVER_CONNSTR";
+
+    /// <summary>The configured connection string, or <see langword="null"/> when the suite is not enabled.</summary>
+    public static string? ConnectionString =>
+        Environment.GetEnvironmentVariable(ConnectionStringVariable);
+
+    /// <summary>Whether the live suite should run.</summary>
+    public static bool IsEnabled => !string.IsNullOrWhiteSpace(ConnectionString);
+
+    /// <summary>Skip reason shown when the suite is not enabled.</summary>
+    public const string SkipReason =
+        "Live SQL Server suite disabled; set AUTH_SQLSERVER_CONNSTR to enable.";
+
+    /// <summary>
+    /// Builds options against a throwaway schema so parallel runs and repeat runs cannot collide,
+    /// and so a failed run leaves no trace in a shared database.
+    /// </summary>
+    /// <param name="schemaName">The unique schema name for this test.</param>
+    /// <returns>Options targeting that schema.</returns>
+    public static SqlServerApiKeyOptions OptionsFor(string schemaName) => new()
+    {
+        ConnectionString = ConnectionString!,
+        Schema = schemaName,
+    };
+
+    /// <summary>Generates a unique, allow-list-legal schema name for one test.</summary>
+    /// <returns>A fresh schema name.</returns>
+    public static string NewSchemaName() => $"zbauth_test_{Guid.NewGuid():N}"[..32];
+
+    /// <summary>Drops the throwaway schema and its three tables.</summary>
+    /// <param name="schemaName">The schema to drop.</param>
+    /// <returns>A task that completes when the schema is gone.</returns>
+    public static async Task DropSchemaAsync(string schemaName)
+    {
+        if (!IsEnabled)
+        {
+            return;
+        }
+
+        await using var connection = new SqlConnection(ConnectionString);
+        await connection.OpenAsync();
+
+        await using SqlCommand command = connection.CreateCommand();
+        command.CommandText = $"""
+            IF OBJECT_ID(N'[{schemaName}].[api_key_audit]', N'U') IS NOT NULL DROP TABLE [{schemaName}].[api_key_audit];
+            IF OBJECT_ID(N'[{schemaName}].[api_keys]', N'U') IS NOT NULL DROP TABLE [{schemaName}].[api_keys];
+            IF OBJECT_ID(N'[{schemaName}].[schema_version]', N'U') IS NOT NULL DROP TABLE [{schemaName}].[schema_version];
+            IF EXISTS (SELECT 1 FROM sys.schemas WHERE name = N'{schemaName}') EXEC(N'DROP SCHEMA [{schemaName}]');
+            """;
+
+        await command.ExecuteNonQueryAsync();
+    }
+
+    /// <summary>Runs a scalar query against the live database.</summary>
+    /// <param name="commandText">The query text.</param>
+    /// <returns>The first column of the first row, or <see langword="null"/>.</returns>
+    public static async Task<object?> ScalarAsync(string commandText)
+    {
+        await using var connection = new SqlConnection(ConnectionString);
+        await connection.OpenAsync();
+
+        await using SqlCommand command = connection.CreateCommand();
+        command.CommandText = commandText;
+
+        object? value = await command.ExecuteScalarAsync();
+        return value == DBNull.Value ? null : value;
+    }
+
+    /// <summary>Runs a non-query statement against the live database.</summary>
+    /// <param name="commandText">The statement text.</param>
+    /// <returns>A task that completes when the statement has run.</returns>
+    public static async Task ExecuteAsync(string commandText)
+    {
+        await using var connection = new SqlConnection(ConnectionString);
+        await connection.OpenAsync();
+
+        await using SqlCommand command = connection.CreateCommand();
+        command.CommandText = commandText;
+
+        await command.ExecuteNonQueryAsync();
+    }
+}
@@ -0,0 +1,499 @@
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests.Live;
+
+/// <summary>
+/// Live contract tests for <see cref="SqlServerApiKeyAdminStore"/> (and the audit store it is
+/// exercised alongside).
+/// </summary>
+/// <remarks>
+/// A one-for-one port of <c>SqliteApiKeyAdminStoreTests</c> — same test names, same assertions — so
+/// the create/revoke/rotate/scope/enable/delete semantics are pinned identically across the two
+/// providers. xUnit constructs a new test-class instance per test, so each test provisions and drops
+/// its own throwaway schema.
+/// </remarks>
+[Collection("live-sqlserver")]
+public sealed class SqlServerApiKeyAdminStoreLiveTests : IAsyncLifetime
+{
+    private readonly string _schema = LiveSqlServer.NewSchemaName();
+
+    private AuthSqlServerConnectionFactory _factory = null!;
+    private SqlServerApiKeyAdminStore _admin = null!;
+    private SqlServerApiKeyStore _read = null!;
+    private SqlServerApiKeyAuditStore _audit = null!;
+
+    /// <inheritdoc />
+    public async Task InitializeAsync()
+    {
+        if (!LiveSqlServer.IsEnabled)
+        {
+            return;
+        }
+
+        _factory = new AuthSqlServerConnectionFactory(LiveSqlServer.OptionsFor(_schema));
+        await new SqlServerApiKeyStoreMigrator(_factory).MigrateAsync(CancellationToken.None);
+        _admin = new SqlServerApiKeyAdminStore(_factory);
+        _read = new SqlServerApiKeyStore(_factory);
+        _audit = new SqlServerApiKeyAuditStore(_factory);
+    }
+
+    /// <inheritdoc />
+    public Task DisposeAsync() => LiveSqlServer.DropSchemaAsync(_schema);
+
+    // --- Create ---
+
+    [SkippableFact]
+    public async Task Create_ThenFindByKeyId_ReturnsRecord()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        ApiKeyRecord record = SampleRecord("key-1");
+
+        await _admin.CreateAsync(record, CancellationToken.None);
+
+        ApiKeyRecord? found = await _read.FindByKeyIdAsync("key-1", CancellationToken.None);
+        Assert.NotNull(found);
+        Assert.Equal(record.SecretHash, found!.SecretHash);
+        Assert.True(record.Scopes.SetEquals(found.Scopes));
+        Assert.Equal(record.ConstraintsJson, found.ConstraintsJson);
+        Assert.Null(found.LastUsedUtc);
+        Assert.Null(found.RevokedUtc);
+    }
+
+    [SkippableFact]
+    public async Task Create_WithExpiresUtc_RoundTripsThroughFindAndList()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        var expiry = new DateTimeOffset(2026, 12, 31, 23, 59, 59, TimeSpan.Zero);
+        ApiKeyRecord record = SampleRecord("key-exp") with { ExpiresUtc = expiry };
+
+        await _admin.CreateAsync(record, CancellationToken.None);
+
+        ApiKeyRecord? found = await _read.FindByKeyIdAsync("key-exp", CancellationToken.None);
+        Assert.Equal(expiry, found!.ExpiresUtc);
+
+        IReadOnlyList<ApiKeyListItem> listed = await _admin.ListAsync(CancellationToken.None);
+        ApiKeyListItem item = Assert.Single(listed, k => k.KeyId == "key-exp");
+        Assert.Equal(expiry, item.ExpiresUtc);
+    }
+
+    [SkippableFact]
+    public async Task Create_WithoutExpiresUtc_ReadsBackAsNull()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(SampleRecord("key-1"), CancellationToken.None);
+
+        ApiKeyRecord? found = await _read.FindByKeyIdAsync("key-1", CancellationToken.None);
+        Assert.Null(found!.ExpiresUtc);
+
+        IReadOnlyList<ApiKeyListItem> listed = await _admin.ListAsync(CancellationToken.None);
+        Assert.Null(Assert.Single(listed, k => k.KeyId == "key-1").ExpiresUtc);
+    }
+
+    // --- Revoke ---
+
+    [SkippableFact]
+    public async Task Revoke_ActiveKey_SetsRevokedAndFindActiveReturnsNull()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(SampleRecord("key-1"), CancellationToken.None);
+        var when = new DateTimeOffset(2026, 5, 31, 9, 0, 0, TimeSpan.Zero);
+
+        bool result = await _admin.RevokeAsync("key-1", when, CancellationToken.None);
+
+        Assert.True(result);
+        Assert.Null(await _read.FindActiveByKeyIdAsync("key-1", CancellationToken.None));
+        ApiKeyRecord? found = await _read.FindByKeyIdAsync("key-1", CancellationToken.None);
+        Assert.Equal(when, found!.RevokedUtc);
+    }
+
+    [SkippableFact]
+    public async Task Revoke_UnknownKey_ReturnsFalse()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        bool result = await _admin.RevokeAsync("missing", DateTimeOffset.UtcNow, CancellationToken.None);
+
+        Assert.False(result);
+    }
+
+    [SkippableFact]
+    public async Task Revoke_AlreadyRevoked_ReturnsFalse()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(SampleRecord("key-1"), CancellationToken.None);
+        await _admin.RevokeAsync("key-1", DateTimeOffset.UtcNow, CancellationToken.None);
+
+        bool result = await _admin.RevokeAsync("key-1", DateTimeOffset.UtcNow, CancellationToken.None);
+
+        Assert.False(result);
+    }
+
+    // --- Rotate ---
+
+    [SkippableFact]
+    public async Task Rotate_ChangesHashAndReactivates()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(SampleRecord("key-1"), CancellationToken.None);
+        await _admin.RevokeAsync("key-1", DateTimeOffset.UtcNow, CancellationToken.None);
+        await _read.MarkUsedAsync("key-1", DateTimeOffset.UtcNow, CancellationToken.None);
+        byte[] newHash = [9, 9, 9, 9];
+
+        bool result = await _admin.RotateAsync("key-1", newHash, CancellationToken.None);
+
+        Assert.True(result);
+        ApiKeyRecord? found = await _read.FindByKeyIdAsync("key-1", CancellationToken.None);
+        Assert.Equal(newHash, found!.SecretHash);
+        Assert.Null(found.RevokedUtc);
+        Assert.Null(found.LastUsedUtc);
+        // Reactivated: now visible as active.
+        Assert.NotNull(await _read.FindActiveByKeyIdAsync("key-1", CancellationToken.None));
+    }
+
+    [SkippableFact]
+    public async Task Rotate_UnknownKey_ReturnsFalse()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        bool result = await _admin.RotateAsync("missing", [1], CancellationToken.None);
+
+        Assert.False(result);
+    }
+
+    // --- SetScopes ---
+
+    [SkippableFact]
+    public async Task SetScopesAsync_ReplacesScopes_AndReturnsTrue()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(
+            SampleRecord("key-1") with { Scopes = new HashSet<string>(["a"], StringComparer.Ordinal) },
+            CancellationToken.None);
+
+        bool result = await _admin.SetScopesAsync(
+            "key-1",
+            new HashSet<string>(["b", "c"], StringComparer.Ordinal),
+            CancellationToken.None);
+
+        Assert.True(result);
+        IReadOnlyList<ApiKeyListItem> listed = await _admin.ListAsync(CancellationToken.None);
+        ApiKeyListItem item = Assert.Single(listed, k => k.KeyId == "key-1");
+        Assert.True(item.Scopes.SetEquals(new HashSet<string>(["b", "c"], StringComparer.Ordinal)));
+    }
+
+    [SkippableFact]
+    public async Task SetScopesAsync_UnknownKey_ReturnsFalse()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        bool result = await _admin.SetScopesAsync(
+            "missing",
+            new HashSet<string>(["b"], StringComparer.Ordinal),
+            CancellationToken.None);
+
+        Assert.False(result);
+    }
+
+    // --- SetEnabled ---
+
+    [SkippableFact]
+    public async Task SetEnabledAsync_False_DisablesKey()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(SampleRecord("key-1"), CancellationToken.None);
+        var when = new DateTimeOffset(2026, 5, 31, 9, 0, 0, TimeSpan.Zero);
+
+        bool result = await _admin.SetEnabledAsync("key-1", enabled: false, when, CancellationToken.None);
+
+        Assert.True(result);
+        Assert.Null(await _read.FindActiveByKeyIdAsync("key-1", CancellationToken.None));
+        ApiKeyRecord? found = await _read.FindByKeyIdAsync("key-1", CancellationToken.None);
+        Assert.Equal(when, found!.RevokedUtc);
+    }
+
+    [SkippableFact]
+    public async Task SetEnabledAsync_True_ReenablesKey_WithoutChangingSecret()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        ApiKeyRecord original = SampleRecord("key-1");
+        await _admin.CreateAsync(original, CancellationToken.None);
+        // Record some usage so we can prove last_used_utc is left untouched on re-enable.
+        var used = new DateTimeOffset(2026, 5, 20, 12, 0, 0, TimeSpan.Zero);
+        await _read.MarkUsedAsync("key-1", used, CancellationToken.None);
+
+        // Disable, then re-enable.
+        await _admin.SetEnabledAsync(
+            "key-1", enabled: false, new DateTimeOffset(2026, 5, 31, 9, 0, 0, TimeSpan.Zero), CancellationToken.None);
+        bool result = await _admin.SetEnabledAsync(
+            "key-1", enabled: true, new DateTimeOffset(2026, 6, 1, 9, 0, 0, TimeSpan.Zero), CancellationToken.None);
+
+        Assert.True(result);
+
+        // Active again, and the secret hash + last-used timestamp are unchanged.
+        ApiKeyRecord? active = await _read.FindActiveByKeyIdAsync("key-1", CancellationToken.None);
+        Assert.NotNull(active);
+        Assert.True(active!.SecretHash.SequenceEqual(original.SecretHash));
+        Assert.Null(active.RevokedUtc);
+        Assert.Equal(used, active.LastUsedUtc);
+    }
+
+    [SkippableFact]
+    public async Task SetEnabledAsync_UnknownKey_ReturnsFalse()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        bool result = await _admin.SetEnabledAsync(
+            "missing", enabled: false, DateTimeOffset.UtcNow, CancellationToken.None);
+
+        Assert.False(result);
+    }
+
+    // --- Delete ---
+
+    [SkippableFact]
+    public async Task Delete_ActiveKey_ReturnsFalseAndKeyStillPresent()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(SampleRecord("key-1"), CancellationToken.None);
+
+        bool result = await _admin.DeleteAsync("key-1", CancellationToken.None);
+
+        Assert.False(result);
+        Assert.NotNull(await _read.FindByKeyIdAsync("key-1", CancellationToken.None));
+    }
+
+    [SkippableFact]
+    public async Task Delete_RevokedKey_ReturnsTrueAndKeyGone()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(SampleRecord("key-1"), CancellationToken.None);
+        await _admin.RevokeAsync("key-1", DateTimeOffset.UtcNow, CancellationToken.None);
+
+        bool result = await _admin.DeleteAsync("key-1", CancellationToken.None);
+
+        Assert.True(result);
+        Assert.Null(await _read.FindByKeyIdAsync("key-1", CancellationToken.None));
+    }
+
+    [SkippableFact]
+    public async Task Delete_UnknownKey_ReturnsFalse()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        bool result = await _admin.DeleteAsync("missing", CancellationToken.None);
+
+        Assert.False(result);
+    }
+
+    // --- keyId guard tests ---
+
+    [SkippableTheory]
+    [InlineData(null)]
+    [InlineData("")]
+    [InlineData("   ")]
+    public async Task Revoke_NullOrWhitespaceKeyId_ThrowsArgumentException(string? keyId)
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        // ArgumentNullException (null) and ArgumentException (empty/whitespace) are both acceptable;
+        // ThrowIfNullOrWhiteSpace throws ArgumentNullException for null, ArgumentException for whitespace.
+        await Assert.ThrowsAnyAsync<ArgumentException>(
+            () => _admin.RevokeAsync(keyId!, DateTimeOffset.UtcNow, CancellationToken.None));
+    }
+
+    [SkippableTheory]
+    [InlineData(null)]
+    [InlineData("")]
+    [InlineData("   ")]
+    public async Task Rotate_NullOrWhitespaceKeyId_ThrowsArgumentException(string? keyId)
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await Assert.ThrowsAnyAsync<ArgumentException>(
+            () => _admin.RotateAsync(keyId!, [1, 2, 3], CancellationToken.None));
+    }
+
+    [SkippableTheory]
+    [InlineData(null)]
+    [InlineData("")]
+    [InlineData("   ")]
+    public async Task Delete_NullOrWhitespaceKeyId_ThrowsArgumentException(string? keyId)
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await Assert.ThrowsAnyAsync<ArgumentException>(
+            () => _admin.DeleteAsync(keyId!, CancellationToken.None));
+    }
+
+    [SkippableTheory]
+    [InlineData(null)]
+    [InlineData("")]
+    [InlineData("   ")]
+    public async Task SetScopesAsync_NullOrWhitespaceKeyId_ThrowsArgumentException(string? keyId)
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await Assert.ThrowsAnyAsync<ArgumentException>(
+            () => _admin.SetScopesAsync(
+                keyId!,
+                new HashSet<string>(["read"], StringComparer.Ordinal),
+                CancellationToken.None));
+    }
+
+    [SkippableTheory]
+    [InlineData(null)]
+    [InlineData("")]
+    [InlineData("   ")]
+    public async Task SetEnabledAsync_NullOrWhitespaceKeyId_ThrowsArgumentException(string? keyId)
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await Assert.ThrowsAnyAsync<ArgumentException>(
+            () => _admin.SetEnabledAsync(keyId!, enabled: false, DateTimeOffset.UtcNow, CancellationToken.None));
+    }
+
+    [SkippableFact]
+    public async Task SetScopesAsync_NullScopes_ThrowsArgumentNullException()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(SampleRecord("key-1"), CancellationToken.None);
+
+        await Assert.ThrowsAsync<ArgumentNullException>(
+            () => _admin.SetScopesAsync("key-1", null!, CancellationToken.None));
+    }
+
+    // --- SetEnabled idempotence ---
+
+    [SkippableFact]
+    public async Task SetEnabledAsync_OnAlreadyActiveKey_ReturnsTrue()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(SampleRecord("key-1"), CancellationToken.None);
+
+        bool result = await _admin.SetEnabledAsync(
+            "key-1", enabled: true, DateTimeOffset.UtcNow, CancellationToken.None);
+
+        Assert.True(result);
+        ApiKeyRecord? active = await _read.FindActiveByKeyIdAsync("key-1", CancellationToken.None);
+        Assert.NotNull(active);
+        Assert.Null(active!.RevokedUtc);
+    }
+
+    [SkippableFact]
+    public async Task SetEnabledAsync_OnAlreadyDisabledKey_OverwritesTimestamp_ReturnsTrue()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _admin.CreateAsync(SampleRecord("key-1"), CancellationToken.None);
+        var t1 = new DateTimeOffset(2026, 5, 1, 10, 0, 0, TimeSpan.Zero);
+        var t2 = new DateTimeOffset(2026, 5, 15, 10, 0, 0, TimeSpan.Zero);
+
+        // Disable at t1.
+        await _admin.SetEnabledAsync("key-1", enabled: false, t1, CancellationToken.None);
+
+        // Disable again at a later t2 (idempotent overwrite — no guard on revoked_utc).
+        bool result = await _admin.SetEnabledAsync("key-1", enabled: false, t2, CancellationToken.None);
+
+        Assert.True(result);
+        IReadOnlyList<ApiKeyListItem> listed = await _admin.ListAsync(CancellationToken.None);
+        ApiKeyListItem item = Assert.Single(listed, k => k.KeyId == "key-1");
+        Assert.Equal(t2, item.RevokedUtc);
+    }
+
+    // --- Audit ---
+
+    [SkippableFact]
+    public async Task Audit_AppendThenListRecent_ReturnsEntry()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        var entry = new ApiKeyAuditEntry(
+            KeyId: "key-1",
+            EventType: "created",
+            RemoteAddress: "10.0.0.1",
+            CreatedUtc: new DateTimeOffset(2026, 5, 31, 10, 0, 0, TimeSpan.Zero),
+            Details: "by admin");
+
+        await _audit.AppendAsync(entry, CancellationToken.None);
+
+        IReadOnlyList<ApiKeyAuditEntry> recent = await _audit.ListRecentAsync(10, CancellationToken.None);
+        Assert.Single(recent);
+        Assert.Equal("key-1", recent[0].KeyId);
+        Assert.Equal("created", recent[0].EventType);
+        Assert.Equal("10.0.0.1", recent[0].RemoteAddress);
+        Assert.Equal("by admin", recent[0].Details);
+    }
+
+    [SkippableFact]
+    public async Task Audit_ListRecent_ReturnsNewestFirst()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _audit.AppendAsync(
+            new ApiKeyAuditEntry("k", "first", null, DateTimeOffset.UtcNow, null), CancellationToken.None);
+        await _audit.AppendAsync(
+            new ApiKeyAuditEntry("k", "second", null, DateTimeOffset.UtcNow, null), CancellationToken.None);
+        await _audit.AppendAsync(
+            new ApiKeyAuditEntry("k", "third", null, DateTimeOffset.UtcNow, null), CancellationToken.None);
+
+        IReadOnlyList<ApiKeyAuditEntry> recent = await _audit.ListRecentAsync(10, CancellationToken.None);
+
+        Assert.Equal(["third", "second", "first"], recent.Select(e => e.EventType));
+    }
+
+    [SkippableFact]
+    public async Task Audit_ListRecent_RespectsLimit()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        for (int i = 0; i < 5; i++)
+        {
+            await _audit.AppendAsync(
+                new ApiKeyAuditEntry("k", $"e{i}", null, DateTimeOffset.UtcNow, null), CancellationToken.None);
+        }
+
+        IReadOnlyList<ApiKeyAuditEntry> recent = await _audit.ListRecentAsync(2, CancellationToken.None);
+
+        Assert.Equal(2, recent.Count);
+        Assert.Equal(["e4", "e3"], recent.Select(e => e.EventType));
+    }
+
+    [SkippableFact]
+    public async Task Audit_NullableFields_RoundTripAsNull()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _audit.AppendAsync(
+            new ApiKeyAuditEntry(null, "anon", null, DateTimeOffset.UtcNow, null), CancellationToken.None);
+
+        IReadOnlyList<ApiKeyAuditEntry> recent = await _audit.ListRecentAsync(10, CancellationToken.None);
+        Assert.Single(recent);
+        Assert.Null(recent[0].KeyId);
+        Assert.Null(recent[0].RemoteAddress);
+        Assert.Null(recent[0].Details);
+    }
+
+    private static ApiKeyRecord SampleRecord(string keyId) => new(
+        KeyId: keyId,
+        KeyPrefix: "mxgw_ab12",
+        SecretHash: [1, 2, 3, 4, 5, 6, 7, 8],
+        DisplayName: "Test Key " + keyId,
+        Scopes: new HashSet<string>(["read", "write"], StringComparer.Ordinal),
+        ConstraintsJson: """{"ipAllow":["10.0.0.0/8"]}""",
+        CreatedUtc: new DateTimeOffset(2026, 5, 1, 8, 30, 0, TimeSpan.Zero),
+        LastUsedUtc: null,
+        RevokedUtc: null);
+}
@@ -0,0 +1,74 @@
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests.Live;
+
+/// <summary>
+/// Live tests for <see cref="SqlServerApiKeyAuditStore"/> that the ported SQLite suite does not
+/// already cover: newest-first ordering under a limit, and the non-positive-limit short circuit.
+/// </summary>
+/// <remarks>
+/// xUnit constructs a new test-class instance per test, so each test provisions and drops its own
+/// throwaway schema.
+/// </remarks>
+[Collection("live-sqlserver")]
+public sealed class SqlServerApiKeyAuditStoreLiveTests : IAsyncLifetime
+{
+    private readonly string _schema = LiveSqlServer.NewSchemaName();
+
+    private SqlServerApiKeyAuditStore _audit = null!;
+
+    /// <inheritdoc />
+    public async Task InitializeAsync()
+    {
+        if (!LiveSqlServer.IsEnabled)
+        {
+            return;
+        }
+
+        var factory = new AuthSqlServerConnectionFactory(LiveSqlServer.OptionsFor(_schema));
+        await new SqlServerApiKeyStoreMigrator(factory).MigrateAsync(CancellationToken.None);
+        _audit = new SqlServerApiKeyAuditStore(factory);
+    }
+
+    /// <inheritdoc />
+    public Task DisposeAsync() => LiveSqlServer.DropSchemaAsync(_schema);
+
+    [SkippableFact]
+    public async Task Append_ThenListRecent_ReturnsNewestFirstWithinLimit()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        // Every row carries the SAME created_utc, so the newest-first order can only come from the
+        // IDENTITY audit_id the ORDER BY names — the timestamp cannot accidentally supply it.
+        var when = new DateTimeOffset(2026, 7, 1, 0, 0, 0, TimeSpan.Zero);
+        for (int i = 0; i < 6; i++)
+        {
+            await _audit.AppendAsync(
+                new ApiKeyAuditEntry("k", $"e{i}", "10.0.0.1", when, $"detail {i}"),
+                CancellationToken.None);
+        }
+
+        IReadOnlyList<ApiKeyAuditEntry> recent = await _audit.ListRecentAsync(3, CancellationToken.None);
+
+        Assert.Equal(3, recent.Count);
+        Assert.Equal(["e5", "e4", "e3"], recent.Select(e => e.EventType));
+        Assert.Equal("detail 5", recent[0].Details);
+        Assert.Equal(when, recent[0].CreatedUtc);
+    }
+
+    [SkippableFact]
+    public async Task ListRecent_NonPositiveLimit_ReturnsEmpty()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await _audit.AppendAsync(
+            new ApiKeyAuditEntry("k", "created", null, DateTimeOffset.UtcNow, null),
+            CancellationToken.None);
+
+        // A non-positive limit short circuits WITHOUT touching the database — SELECT TOP (0) would
+        // be legal but SELECT TOP (-1) is a syntax error, so the guard is load-bearing here in a way
+        // it is not for SQLite's LIMIT.
+        Assert.Empty(await _audit.ListRecentAsync(0, CancellationToken.None));
+        Assert.Empty(await _audit.ListRecentAsync(-5, CancellationToken.None));
+    }
+}
@@ -0,0 +1,340 @@
+using System.Data;
+using Microsoft.Data.SqlClient;
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+using ZB.MOM.WW.Auth.ApiKeys.Sqlite;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests.Live;
+
+/// <summary>
+/// Live contract tests for <see cref="SqlServerApiKeyStore"/>.
+/// </summary>
+/// <remarks>
+/// A one-for-one port of <c>SqliteApiKeyStoreTests</c> (same test names, same assertions) so the two
+/// providers are pinned to identical semantics: whichever store a host composes, a row means the
+/// same thing. xUnit constructs a new test-class instance per test, so each test provisions and
+/// drops its own throwaway schema — the isolation the SQLite suite gets from a per-test temp file.
+/// </remarks>
+[Collection("live-sqlserver")]
+public sealed class SqlServerApiKeyStoreLiveTests : IAsyncLifetime
+{
+    private readonly string _schema = LiveSqlServer.NewSchemaName();
+
+    private AuthSqlServerConnectionFactory _factory = null!;
+    private SqlServerApiKeyStore _store = null!;
+
+    /// <inheritdoc />
+    public async Task InitializeAsync()
+    {
+        if (!LiveSqlServer.IsEnabled)
+        {
+            return;
+        }
+
+        _factory = new AuthSqlServerConnectionFactory(LiveSqlServer.OptionsFor(_schema));
+        await new SqlServerApiKeyStoreMigrator(_factory).MigrateAsync(CancellationToken.None);
+        _store = new SqlServerApiKeyStore(_factory);
+    }
+
+    /// <inheritdoc />
+    public Task DisposeAsync() => LiveSqlServer.DropSchemaAsync(_schema);
+
+    [SkippableFact]
+    public async Task FindByKeyId_AfterInsert_ReturnsEqualRecord()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        ApiKeyRecord record = SampleRecord("key-1");
+        await InsertAsync(record);
+
+        ApiKeyRecord? found = await _store.FindByKeyIdAsync("key-1", CancellationToken.None);
+
+        Assert.NotNull(found);
+        AssertRecordEqual(record, found!);
+    }
+
+    [SkippableFact]
+    public async Task FindByKeyId_ReturnsRevokedRecord()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        ApiKeyRecord record = SampleRecord("key-revoked") with
+        {
+            RevokedUtc = new DateTimeOffset(2026, 1, 2, 3, 4, 5, TimeSpan.Zero),
+        };
+        await InsertAsync(record);
+
+        ApiKeyRecord? found = await _store.FindByKeyIdAsync("key-revoked", CancellationToken.None);
+
+        Assert.NotNull(found);
+        Assert.NotNull(found!.RevokedUtc);
+    }
+
+    [SkippableFact]
+    public async Task FindActiveByKeyId_RevokedKey_ReturnsNull()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        ApiKeyRecord record = SampleRecord("key-revoked") with
+        {
+            RevokedUtc = new DateTimeOffset(2026, 1, 2, 3, 4, 5, TimeSpan.Zero),
+        };
+        await InsertAsync(record);
+
+        ApiKeyRecord? found = await _store.FindActiveByKeyIdAsync("key-revoked", CancellationToken.None);
+
+        Assert.Null(found);
+    }
+
+    [SkippableFact]
+    public async Task FindActiveByKeyId_ActiveKey_ReturnsRecord()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await InsertAsync(SampleRecord("key-active"));
+
+        ApiKeyRecord? found = await _store.FindActiveByKeyIdAsync("key-active", CancellationToken.None);
+
+        Assert.NotNull(found);
+    }
+
+    [SkippableFact]
+    public async Task FindByKeyId_UnknownKey_ReturnsNull()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        ApiKeyRecord? found = await _store.FindByKeyIdAsync("missing", CancellationToken.None);
+
+        Assert.Null(found);
+    }
+
+    [SkippableFact]
+    public async Task MarkUsed_ActiveKey_UpdatesLastUsed()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await InsertAsync(SampleRecord("key-active"));
+        var when = new DateTimeOffset(2026, 5, 31, 12, 0, 0, TimeSpan.Zero);
+
+        await _store.MarkUsedAsync("key-active", when, CancellationToken.None);
+
+        ApiKeyRecord? found = await _store.FindByKeyIdAsync("key-active", CancellationToken.None);
+        Assert.Equal(when, found!.LastUsedUtc);
+    }
+
+    [SkippableFact]
+    public async Task MarkUsed_RevokedKey_DoesNotUpdateLastUsed()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        ApiKeyRecord record = SampleRecord("key-revoked") with
+        {
+            RevokedUtc = new DateTimeOffset(2026, 1, 2, 3, 4, 5, TimeSpan.Zero),
+        };
+        await InsertAsync(record);
+        var when = new DateTimeOffset(2026, 5, 31, 12, 0, 0, TimeSpan.Zero);
+
+        await _store.MarkUsedAsync("key-revoked", when, CancellationToken.None);
+
+        ApiKeyRecord? found = await _store.FindByKeyIdAsync("key-revoked", CancellationToken.None);
+        Assert.Null(found!.LastUsedUtc);
+    }
+
+    // --- SQL-Server-specific: DATETIMEOFFSET(7) must not truncate to milliseconds ---
+
+    [SkippableFact]
+    public async Task FindByKeyId_RoundTripsDateTimeOffsetToTheTick()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        // Sub-millisecond ticks: SQLite persists round-trip ("O") text and keeps them for free,
+        // whereas SQL Server keeps them only because the columns are DATETIMEOFFSET(7) and the
+        // parameters are bound as SqlDbType.DateTimeOffset. A DATETIME/DATETIME2(3) column or an
+        // untyped parameter would silently round here.
+        var created = new DateTimeOffset(2026, 3, 4, 5, 6, 7, TimeSpan.Zero).AddTicks(1234567);
+        var expires = new DateTimeOffset(2027, 3, 4, 5, 6, 7, TimeSpan.Zero).AddTicks(7654321);
+        ApiKeyRecord record = SampleRecord("key-ticks") with { CreatedUtc = created, ExpiresUtc = expires };
+        await InsertAsync(record);
+
+        ApiKeyRecord? found = await _store.FindByKeyIdAsync("key-ticks", CancellationToken.None);
+
+        Assert.NotNull(found);
+        Assert.Equal(created, found!.CreatedUtc);
+        Assert.Equal(created.Ticks, found.CreatedUtc.Ticks);
+        Assert.Equal(expires, found.ExpiresUtc);
+        Assert.Equal(expires.Ticks, found.ExpiresUtc!.Value.Ticks);
+    }
+
+    // --- keyId guard tests ---
+
+    [SkippableTheory]
+    [InlineData(null)]
+    [InlineData("")]
+    [InlineData("   ")]
+    public async Task FindByKeyId_NullOrWhitespaceKeyId_ThrowsArgumentException(string? keyId)
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        // ArgumentNullException (null) and ArgumentException (empty/whitespace) are both acceptable;
+        // ThrowIfNullOrWhiteSpace throws ArgumentNullException for null, ArgumentException for whitespace.
+        await Assert.ThrowsAnyAsync<ArgumentException>(
+            () => _store.FindByKeyIdAsync(keyId!, CancellationToken.None));
+    }
+
+    [SkippableTheory]
+    [InlineData(null)]
+    [InlineData("")]
+    [InlineData("   ")]
+    public async Task FindActiveByKeyId_NullOrWhitespaceKeyId_ThrowsArgumentException(string? keyId)
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await Assert.ThrowsAnyAsync<ArgumentException>(
+            () => _store.FindActiveByKeyIdAsync(keyId!, CancellationToken.None));
+    }
+
+    [SkippableTheory]
+    [InlineData(null)]
+    [InlineData("")]
+    [InlineData("   ")]
+    public async Task MarkUsed_NullOrWhitespaceKeyId_ThrowsArgumentException(string? keyId)
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await Assert.ThrowsAnyAsync<ArgumentException>(
+            () => _store.MarkUsedAsync(keyId!, DateTimeOffset.UtcNow, CancellationToken.None));
+    }
+
+    [SkippableFact]
+    public void ScopeSerializer_RoundTripsAndSortsOrdinally()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        var unsorted = new HashSet<string>(["zeta", "alpha", "mike"], StringComparer.Ordinal);
+        var differentOrder = new HashSet<string>(["mike", "zeta", "alpha"], StringComparer.Ordinal);
+
+        string a = ScopeSerializer.Serialize(unsorted);
+        string b = ScopeSerializer.Serialize(differentOrder);
+
+        // Equal sets must produce identical column text regardless of insertion order.
+        Assert.Equal(a, b);
+        Assert.Equal("""["alpha","mike","zeta"]""", a);
+
+        IReadOnlySet<string> roundTripped = ScopeSerializer.Deserialize(a);
+        Assert.True(roundTripped.SetEquals(unsorted));
+    }
+
+    [SkippableFact]
+    public void ScopeSerializer_DeserializeNullOrEmpty_ReturnsEmptySet()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        Assert.Empty(ScopeSerializer.Deserialize(null));
+        Assert.Empty(ScopeSerializer.Deserialize(""));
+    }
+
+    // --- Auth-003: corrupt scopes JSON must fail closed (empty set), never throw JsonException ---
+
+    [SkippableTheory]
+    [InlineData("not json at all")]
+    [InlineData("{")]
+    [InlineData("{\"a\":1}")]   // valid JSON, but an object, not a string[]
+    [InlineData("42")]          // valid JSON, but a number
+    [InlineData("[\"read\",")]  // truncated/partial write
+    public void ScopeSerializer_DeserializeMalformed_ReturnsEmptySet_DoesNotThrow(string value)
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        // A poisoned scopes column (tampering, partial write, format change, buggy writer) must
+        // degrade to a zero-scope set rather than throwing on the verification hot path.
+        IReadOnlySet<string> scopes = ScopeSerializer.Deserialize(value);
+        Assert.Empty(scopes);
+    }
+
+    [SkippableFact]
+    public async Task FindByKeyId_CorruptScopesColumn_ReturnsRecordWithEmptyScopes_DoesNotThrow()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        // Insert a row whose scopes column holds malformed (non-array) JSON, then read it through
+        // the store. The store must NOT propagate a JsonException out of FindByKeyIdAsync (which the
+        // verifier relies on for its "only exception path is cancellation" contract).
+        await InsertWithRawScopesAsync("key-corrupt", scopesJson: "{ this is not valid json");
+
+        ApiKeyRecord? found = await _store.FindByKeyIdAsync("key-corrupt", CancellationToken.None);
+
+        Assert.NotNull(found);
+        Assert.Empty(found!.Scopes);
+    }
+
+    private static ApiKeyRecord SampleRecord(string keyId) => new(
+        KeyId: keyId,
+        KeyPrefix: "mxgw_ab12",
+        SecretHash: [1, 2, 3, 4, 5, 6, 7, 8],
+        DisplayName: "Test Key " + keyId,
+        Scopes: new HashSet<string>(["read", "write"], StringComparer.Ordinal),
+        ConstraintsJson: """{"ipAllow":["10.0.0.0/8"]}""",
+        CreatedUtc: new DateTimeOffset(2026, 5, 1, 8, 30, 0, TimeSpan.Zero),
+        LastUsedUtc: null,
+        RevokedUtc: null);
+
+    private static void AssertRecordEqual(ApiKeyRecord expected, ApiKeyRecord actual)
+    {
+        Assert.Equal(expected.KeyId, actual.KeyId);
+        Assert.Equal(expected.KeyPrefix, actual.KeyPrefix);
+        Assert.Equal(expected.SecretHash, actual.SecretHash);
+        Assert.Equal(expected.DisplayName, actual.DisplayName);
+        Assert.True(expected.Scopes.SetEquals(actual.Scopes));
+        Assert.Equal(expected.ConstraintsJson, actual.ConstraintsJson);
+        Assert.Equal(expected.CreatedUtc, actual.CreatedUtc);
+        Assert.Equal(expected.LastUsedUtc, actual.LastUsedUtc);
+        Assert.Equal(expected.RevokedUtc, actual.RevokedUtc);
+    }
+
+    private async Task InsertAsync(ApiKeyRecord record)
+    {
+        await using SqlConnection connection =
+            await _factory.OpenConnectionAsync(CancellationToken.None);
+        await using SqlCommand command = _factory.CreateCommand(connection, $"""
+            INSERT INTO [{_schema}].[{SqlServerAuthSchema.ApiKeysTable}] (
+                key_id, key_prefix, secret_hash, display_name, scopes,
+                constraints, created_utc, last_used_utc, revoked_utc, expires_utc)
+            VALUES (
+                @key_id, @key_prefix, @secret_hash, @display_name, @scopes,
+                @constraints, @created_utc, @last_used_utc, @revoked_utc, @expires_utc);
+            """);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = record.KeyId;
+        command.Parameters.Add("@key_prefix", SqlDbType.NVarChar, 16).Value = record.KeyPrefix;
+        command.Parameters.Add("@secret_hash", SqlDbType.VarBinary, 64).Value = record.SecretHash;
+        command.Parameters.Add("@display_name", SqlDbType.NVarChar, 256).Value = record.DisplayName;
+        command.Parameters.Add("@scopes", SqlDbType.NVarChar, -1).Value = ScopeSerializer.Serialize(record.Scopes);
+        command.Parameters.Add("@constraints", SqlDbType.NVarChar, -1).Value = (object?)record.ConstraintsJson ?? DBNull.Value;
+        command.Parameters.Add("@created_utc", SqlDbType.DateTimeOffset).Value = record.CreatedUtc;
+        command.Parameters.Add("@last_used_utc", SqlDbType.DateTimeOffset).Value = (object?)record.LastUsedUtc ?? DBNull.Value;
+        command.Parameters.Add("@revoked_utc", SqlDbType.DateTimeOffset).Value = (object?)record.RevokedUtc ?? DBNull.Value;
+        command.Parameters.Add("@expires_utc", SqlDbType.DateTimeOffset).Value = (object?)record.ExpiresUtc ?? DBNull.Value;
+        await command.ExecuteNonQueryAsync(CancellationToken.None);
+    }
+
+    private async Task InsertWithRawScopesAsync(string keyId, string scopesJson)
+    {
+        // Writes the scopes column verbatim (NOT via ScopeSerializer.Serialize) so a malformed
+        // value can be persisted to simulate tampering / a partial or buggy write.
+        await using SqlConnection connection =
+            await _factory.OpenConnectionAsync(CancellationToken.None);
+        await using SqlCommand command = _factory.CreateCommand(connection, $"""
+            INSERT INTO [{_schema}].[{SqlServerAuthSchema.ApiKeysTable}] (
+                key_id, key_prefix, secret_hash, display_name, scopes,
+                constraints, created_utc, last_used_utc, revoked_utc, expires_utc)
+            VALUES (
+                @key_id, @key_prefix, @secret_hash, @display_name, @scopes,
+                NULL, @created_utc, NULL, NULL, NULL);
+            """);
+        command.Parameters.Add("@key_id", SqlDbType.NVarChar, 64).Value = keyId;
+        command.Parameters.Add("@key_prefix", SqlDbType.NVarChar, 16).Value = "mxgw";
+        command.Parameters.Add("@secret_hash", SqlDbType.VarBinary, 64).Value = new byte[] { 1, 2, 3 };
+        command.Parameters.Add("@display_name", SqlDbType.NVarChar, 256).Value = "Corrupt Key";
+        command.Parameters.Add("@scopes", SqlDbType.NVarChar, -1).Value = scopesJson;
+        command.Parameters.Add("@created_utc", SqlDbType.DateTimeOffset).Value = DateTimeOffset.UnixEpoch;
+        await command.ExecuteNonQueryAsync(CancellationToken.None);
+    }
+}
@@ -0,0 +1,145 @@
+using System.Globalization;
+using ZB.MOM.WW.Auth.ApiKeys.Sqlite;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests.Live;
+
+/// <summary>
+/// Live provisioning tests for <see cref="SqlServerApiKeyStoreMigrator"/>.
+/// </summary>
+/// <remarks>
+/// Every test provisions its own throwaway schema — the concurrency case in particular needs a
+/// virgin one, because the race the retry loop exists for only happens while the schema and tables
+/// do not yet exist. All schemas created by the class are dropped in <see cref="DisposeAsync"/>.
+/// </remarks>
+[Collection("live-sqlserver")]
+public sealed class SqlServerApiKeyStoreMigratorLiveTests : IAsyncLifetime
+{
+    private readonly List<string> _schemas = [];
+
+    /// <inheritdoc />
+    public Task InitializeAsync() => Task.CompletedTask;
+
+    /// <inheritdoc />
+    public async Task DisposeAsync()
+    {
+        foreach (string schema in _schemas)
+        {
+            await LiveSqlServer.DropSchemaAsync(schema);
+        }
+    }
+
+    private string NewSchema()
+    {
+        string schema = LiveSqlServer.NewSchemaName();
+        _schemas.Add(schema);
+        return schema;
+    }
+
+    private static SqlServerApiKeyStoreMigrator MigratorFor(string schema) =>
+        new(new AuthSqlServerConnectionFactory(LiveSqlServer.OptionsFor(schema)));
+
+    [SkippableFact]
+    public async Task Migrate_OnEmptyDatabase_CreatesSchemaTablesAndVersionRow()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        string schema = NewSchema();
+
+        await MigratorFor(schema).MigrateAsync(CancellationToken.None);
+
+        Assert.Equal(
+            1,
+            Convert.ToInt32(
+                await LiveSqlServer.ScalarAsync(
+                    $"SELECT COUNT(*) FROM sys.schemas WHERE name = N'{schema}'"),
+                CultureInfo.InvariantCulture));
+
+        foreach (string table in new[]
+        {
+            SqlServerAuthSchema.SchemaVersionTable,
+            SqlServerAuthSchema.ApiKeysTable,
+            SqlServerAuthSchema.ApiKeyAuditTable,
+        })
+        {
+            Assert.NotNull(
+                await LiveSqlServer.ScalarAsync($"SELECT OBJECT_ID(N'[{schema}].[{table}]', N'U')"));
+        }
+
+        Assert.Equal(
+            SqlServerAuthSchema.CurrentVersion,
+            Convert.ToInt32(
+                await LiveSqlServer.ScalarAsync(
+                    $"SELECT version FROM [{schema}].[{SqlServerAuthSchema.SchemaVersionTable}] WHERE id = 1"),
+                CultureInfo.InvariantCulture));
+    }
+
+    [SkippableFact]
+    public async Task Migrate_Twice_IsIdempotent()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        string schema = NewSchema();
+        SqlServerApiKeyStoreMigrator migrator = MigratorFor(schema);
+
+        await migrator.MigrateAsync(CancellationToken.None);
+        await migrator.MigrateAsync(CancellationToken.None);
+
+        Assert.Equal(
+            1,
+            Convert.ToInt32(
+                await LiveSqlServer.ScalarAsync(
+                    $"SELECT COUNT(*) FROM [{schema}].[{SqlServerAuthSchema.SchemaVersionTable}]"),
+                CultureInfo.InvariantCulture));
+        Assert.Equal(
+            SqlServerAuthSchema.CurrentVersion,
+            Convert.ToInt32(
+                await LiveSqlServer.ScalarAsync(
+                    $"SELECT version FROM [{schema}].[{SqlServerAuthSchema.SchemaVersionTable}] WHERE id = 1"),
+                CultureInfo.InvariantCulture));
+    }
+
+    [SkippableFact]
+    public async Task Migrate_Concurrently_FromEightTasks_AllSucceed()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        // A virgin schema provisioned by eight nodes at once — the central-pair first boot, scaled
+        // up. Each task gets its own factory (and therefore its own connection) so the DDL really
+        // does run on eight sessions.
+        string schema = NewSchema();
+
+        await Task.WhenAll(Enumerable.Range(0, 8).Select(_ =>
+            Task.Run(() => MigratorFor(schema).MigrateAsync(CancellationToken.None))));
+
+        Assert.Equal(
+            1,
+            Convert.ToInt32(
+                await LiveSqlServer.ScalarAsync(
+                    $"SELECT COUNT(*) FROM [{schema}].[{SqlServerAuthSchema.SchemaVersionTable}]"),
+                CultureInfo.InvariantCulture));
+        Assert.Equal(
+            SqlServerAuthSchema.CurrentVersion,
+            Convert.ToInt32(
+                await LiveSqlServer.ScalarAsync(
+                    $"SELECT version FROM [{schema}].[{SqlServerAuthSchema.SchemaVersionTable}] WHERE id = 1"),
+                CultureInfo.InvariantCulture));
+    }
+
+    [SkippableFact]
+    public async Task Migrate_WhenStoredVersionIsNewer_Throws()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        string schema = NewSchema();
+        SqlServerApiKeyStoreMigrator migrator = MigratorFor(schema);
+
+        await migrator.MigrateAsync(CancellationToken.None);
+        await LiveSqlServer.ExecuteAsync(
+            $"UPDATE [{schema}].[{SqlServerAuthSchema.SchemaVersionTable}] SET version = 99 WHERE id = 1");
+
+        AuthStoreMigrationException exception = await Assert.ThrowsAsync<AuthStoreMigrationException>(
+            () => migrator.MigrateAsync(CancellationToken.None));
+
+        Assert.Contains("99", exception.Message, StringComparison.Ordinal);
+    }
+}
@@ -0,0 +1,283 @@
+using Microsoft.Data.Sqlite;
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+using ZB.MOM.WW.Auth.ApiKeys.Sqlite;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests.Live;
+
+/// <summary>
+/// Live tests for <see cref="SqliteApiKeyStoreImporter"/> — the one-shot per-node SQLite to shared
+/// SQL Server copy.
+/// </summary>
+/// <remarks>
+/// Each test owns a throwaway SQL Server schema AND a throwaway SQLite file under the temp
+/// directory; both are torn down in <see cref="DisposeAsync"/>, so a failed run leaves nothing
+/// behind on a shared database or the developer's disk.
+/// </remarks>
+[Collection("live-sqlserver")]
+public sealed class SqliteApiKeyStoreImporterLiveTests : IAsyncLifetime
+{
+    private readonly string _schema = LiveSqlServer.NewSchemaName();
+    private readonly string _sqlitePath =
+        Path.Combine(Path.GetTempPath(), $"zbauth-import-{Guid.NewGuid():N}.db");
+
+    private AuthSqlServerConnectionFactory _factory = null!;
+    private SqlServerApiKeyAdminStore _sqlAdmin = null!;
+    private SqlServerApiKeyStore _sqlRead = null!;
+    private SqlServerApiKeyAuditStore _sqlAudit = null!;
+    private SqliteApiKeyStoreImporter _importer = null!;
+
+    /// <inheritdoc />
+    public async Task InitializeAsync()
+    {
+        if (!LiveSqlServer.IsEnabled)
+        {
+            return;
+        }
+
+        _factory = new AuthSqlServerConnectionFactory(LiveSqlServer.OptionsFor(_schema));
+        await new SqlServerApiKeyStoreMigrator(_factory).MigrateAsync(CancellationToken.None);
+        _sqlAdmin = new SqlServerApiKeyAdminStore(_factory);
+        _sqlRead = new SqlServerApiKeyStore(_factory);
+        _sqlAudit = new SqlServerApiKeyAuditStore(_factory);
+        _importer = new SqliteApiKeyStoreImporter(_factory);
+    }
+
+    /// <inheritdoc />
+    public async Task DisposeAsync()
+    {
+        // SQLite pools connections by connection string; the handle must be released before the
+        // file can be removed cleanly.
+        SqliteConnection.ClearAllPools();
+
+        foreach (string path in new[] { _sqlitePath, _sqlitePath + "-wal", _sqlitePath + "-shm" })
+        {
+            if (File.Exists(path))
+            {
+                File.Delete(path);
+            }
+        }
+
+        await LiveSqlServer.DropSchemaAsync(_schema);
+    }
+
+    [SkippableFact]
+    public async Task Import_CopiesKeysAndAuditRows_AndReturnsCounts()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        (ApiKeyRecord keyA, ApiKeyRecord keyB) = await SeedSqliteAsync();
+
+        ApiKeyImportResult result = await _importer.ImportAsync(_sqlitePath, CancellationToken.None);
+
+        Assert.True(result.SourceFound);
+        Assert.Equal(2, result.KeysInserted);
+        Assert.Equal(0, result.KeysSkipped);
+        Assert.Equal(3, result.AuditInserted);
+        Assert.Equal(0, result.AuditSkipped);
+
+        await AssertRecordMatchesAsync(keyA);
+        await AssertRecordMatchesAsync(keyB);
+
+        IReadOnlyList<ApiKeyAuditEntry> audit = await _sqlAudit.ListRecentAsync(10, CancellationToken.None);
+        Assert.Equal(3, audit.Count);
+        Assert.Contains(audit, e => e.KeyId is null && e.EventType == "auth_failed");
+        Assert.Contains(audit, e => e.KeyId == "key-a" && e.EventType == "created");
+        Assert.Contains(audit, e => e.KeyId == "key-b" && e.EventType == "created");
+    }
+
+    [SkippableFact]
+    public async Task Import_Twice_IsIdempotent()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await SeedSqliteAsync();
+
+        await _importer.ImportAsync(_sqlitePath, CancellationToken.None);
+        ApiKeyImportResult second = await _importer.ImportAsync(_sqlitePath, CancellationToken.None);
+
+        Assert.True(second.SourceFound);
+        Assert.Equal(0, second.KeysInserted);
+        Assert.Equal(2, second.KeysSkipped);
+        Assert.Equal(0, second.AuditInserted);
+        Assert.Equal(3, second.AuditSkipped);
+
+        Assert.Equal(2, (await _sqlAdmin.ListAsync(CancellationToken.None)).Count);
+        Assert.Equal(3, (await _sqlAudit.ListRecentAsync(10, CancellationToken.None)).Count);
+    }
+
+    [SkippableFact]
+    public async Task Import_UnionsWithExistingRows()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        var lastUsed = new DateTimeOffset(2026, 7, 4, 12, 0, 0, TimeSpan.Zero);
+        ApiKeyRecord keyC = SampleRecord("key-c") with { LastUsedUtc = lastUsed };
+        await _sqlAdmin.CreateAsync(keyC, CancellationToken.None);
+
+        await SeedSqliteAsync();
+
+        ApiKeyImportResult result = await _importer.ImportAsync(_sqlitePath, CancellationToken.None);
+
+        Assert.Equal(2, result.KeysInserted);
+        Assert.Equal(0, result.KeysSkipped);
+
+        IReadOnlyList<ApiKeyListItem> listed = await _sqlAdmin.ListAsync(CancellationToken.None);
+        Assert.Equal(
+            ["key-a", "key-b", "key-c"],
+            listed.Select(k => k.KeyId).OrderBy(k => k, StringComparer.Ordinal));
+
+        // The pre-existing row is untouched — the importer only ever inserts.
+        Assert.Equal(lastUsed, Assert.Single(listed, k => k.KeyId == "key-c").LastUsedUtc);
+    }
+
+    [SkippableFact]
+    public async Task Import_MissingFile_ReturnsZeroWithoutThrowing()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        ApiKeyImportResult result = await _importer.ImportAsync(
+            Path.Combine(Path.GetTempPath(), $"zbauth-absent-{Guid.NewGuid():N}.db"),
+            CancellationToken.None);
+
+        Assert.False(result.SourceFound);
+        Assert.Equal(0, result.KeysInserted);
+        Assert.Equal(0, result.KeysSkipped);
+        Assert.Equal(0, result.AuditInserted);
+        Assert.Equal(0, result.AuditSkipped);
+    }
+
+    [SkippableFact]
+    public async Task Import_PreV3FileWithoutExpiresColumn_ReadsExpiryAsNull()
+    {
+        Skip.IfNot(LiveSqlServer.IsEnabled, LiveSqlServer.SkipReason);
+
+        await CreatePreV3SqliteFileAsync();
+
+        ApiKeyImportResult result = await _importer.ImportAsync(_sqlitePath, CancellationToken.None);
+
+        Assert.True(result.SourceFound);
+        Assert.Equal(1, result.KeysInserted);
+
+        ApiKeyRecord? found = await _sqlRead.FindByKeyIdAsync("key-old", CancellationToken.None);
+        Assert.NotNull(found);
+        Assert.Null(found!.ExpiresUtc);
+        Assert.Equal("legacy key", found.DisplayName);
+        Assert.Equal(new DateTimeOffset(2026, 1, 2, 3, 4, 5, TimeSpan.Zero), found.CreatedUtc);
+    }
+
+    // --- helpers ---
+
+    /// <summary>
+    /// Seeds the throwaway SQLite file with two keys (one revoked, one with an expiry) and three
+    /// audit rows (one of them key-less), then returns the two key records as SQLite holds them.
+    /// </summary>
+    private async Task<(ApiKeyRecord KeyA, ApiKeyRecord KeyB)> SeedSqliteAsync()
+    {
+        var sqliteFactory = new AuthSqliteConnectionFactory(_sqlitePath);
+        await new SqliteAuthStoreMigrator(sqliteFactory).MigrateAsync(CancellationToken.None);
+
+        var admin = new SqliteApiKeyAdminStore(sqliteFactory);
+        var audit = new SqliteApiKeyAuditStore(sqliteFactory);
+        var read = new SqliteApiKeyStore(sqliteFactory);
+
+        ApiKeyRecord keyA = SampleRecord("key-a");
+        ApiKeyRecord keyB = SampleRecord("key-b") with
+        {
+            ExpiresUtc = new DateTimeOffset(2027, 3, 1, 0, 0, 0, TimeSpan.Zero),
+        };
+
+        await admin.CreateAsync(keyA, CancellationToken.None);
+        await admin.CreateAsync(keyB, CancellationToken.None);
+        await admin.RevokeAsync("key-a", new DateTimeOffset(2026, 6, 1, 7, 8, 9, TimeSpan.Zero), CancellationToken.None);
+
+        await audit.AppendAsync(
+            new ApiKeyAuditEntry("key-a", "created", "10.0.0.1", new DateTimeOffset(2026, 5, 1, 8, 30, 0, TimeSpan.Zero), "by admin"),
+            CancellationToken.None);
+        await audit.AppendAsync(
+            new ApiKeyAuditEntry("key-b", "created", null, new DateTimeOffset(2026, 5, 1, 8, 31, 0, TimeSpan.Zero), null),
+            CancellationToken.None);
+        await audit.AppendAsync(
+            new ApiKeyAuditEntry(null, "auth_failed", "10.0.0.9", new DateTimeOffset(2026, 5, 2, 9, 0, 0, TimeSpan.Zero), "unknown key"),
+            CancellationToken.None);
+
+        // Re-read so the assertions compare against exactly what SQLite holds, revocation included.
+        ApiKeyRecord storedA = (await read.FindByKeyIdAsync("key-a", CancellationToken.None))!;
+        ApiKeyRecord storedB = (await read.FindByKeyIdAsync("key-b", CancellationToken.None))!;
+
+        return (storedA, storedB);
+    }
+
+    /// <summary>
+    /// Writes a v2-shaped SQLite file by hand — <c>api_keys</c> WITHOUT <c>expires_utc</c> — and
+    /// inserts one key row, reproducing a store created before schema v3.
+    /// </summary>
+    private async Task CreatePreV3SqliteFileAsync()
+    {
+        await using var connection = new SqliteConnection(
+            new SqliteConnectionStringBuilder { DataSource = _sqlitePath }.ToString());
+        await connection.OpenAsync();
+
+        await using SqliteCommand ddl = connection.CreateCommand();
+        ddl.CommandText = """
+            CREATE TABLE api_keys (
+                key_id TEXT PRIMARY KEY,
+                key_prefix TEXT NOT NULL,
+                secret_hash BLOB NOT NULL,
+                display_name TEXT NOT NULL,
+                scopes TEXT NOT NULL,
+                constraints TEXT NULL,
+                created_utc TEXT NOT NULL,
+                last_used_utc TEXT NULL,
+                revoked_utc TEXT NULL
+            );
+            CREATE TABLE api_key_audit (
+                audit_id INTEGER PRIMARY KEY AUTOINCREMENT,
+                key_id TEXT NULL,
+                event_type TEXT NOT NULL,
+                remote_address TEXT NULL,
+                created_utc TEXT NOT NULL,
+                details TEXT NULL
+            );
+            """;
+        await ddl.ExecuteNonQueryAsync();
+
+        await using SqliteCommand insert = connection.CreateCommand();
+        insert.CommandText = """
+            INSERT INTO api_keys (
+                key_id, key_prefix, secret_hash, display_name, scopes,
+                constraints, created_utc, last_used_utc, revoked_utc)
+            VALUES ('key-old', 'sbk_old1', $hash, 'legacy key', 'read', NULL, $created, NULL, NULL);
+            """;
+        insert.Parameters.Add("$hash", SqliteType.Blob).Value = new byte[] { 4, 5, 6 };
+        insert.Parameters.AddWithValue(
+            "$created", new DateTimeOffset(2026, 1, 2, 3, 4, 5, TimeSpan.Zero).ToString("O"));
+        await insert.ExecuteNonQueryAsync();
+    }
+
+    private async Task AssertRecordMatchesAsync(ApiKeyRecord expected)
+    {
+        ApiKeyRecord? actual = await _sqlRead.FindByKeyIdAsync(expected.KeyId, CancellationToken.None);
+
+        Assert.NotNull(actual);
+        Assert.Equal(expected.KeyPrefix, actual!.KeyPrefix);
+        Assert.True(expected.SecretHash.SequenceEqual(actual.SecretHash));
+        Assert.Equal(expected.DisplayName, actual.DisplayName);
+        Assert.True(expected.Scopes.SetEquals(actual.Scopes));
+        Assert.Equal(expected.ConstraintsJson, actual.ConstraintsJson);
+        Assert.Equal(expected.CreatedUtc, actual.CreatedUtc);
+        Assert.Equal(expected.LastUsedUtc, actual.LastUsedUtc);
+        Assert.Equal(expected.RevokedUtc, actual.RevokedUtc);
+        Assert.Equal(expected.ExpiresUtc, actual.ExpiresUtc);
+    }
+
+    private static ApiKeyRecord SampleRecord(string keyId) => new(
+        KeyId: keyId,
+        KeyPrefix: "sbk_ab12",
+        SecretHash: [1, 2, 3, 4, 5, 6, 7, 8],
+        DisplayName: "Test Key " + keyId,
+        Scopes: new HashSet<string>(["read", "write"], StringComparer.Ordinal),
+        ConstraintsJson: """{"ipAllow":["10.0.0.0/8"]}""",
+        CreatedUtc: new DateTimeOffset(2026, 5, 1, 8, 30, 0, TimeSpan.Zero),
+        LastUsedUtc: null,
+        RevokedUtc: null);
+}
@@ -0,0 +1,16 @@
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests;
+
+public sealed class SqlServerApiKeyOptionsTests
+{
+    [Fact]
+    public void Validate_RejectsMissingConnectionString() =>
+        Assert.Throws<InvalidOperationException>(() => new SqlServerApiKeyOptions { ConnectionString = " " }.Validate());
+
+    [Fact]
+    public void Validate_RejectsUnsafeSchemaName() =>
+        Assert.Throws<InvalidOperationException>(() => new SqlServerApiKeyOptions { ConnectionString = "Server=x", Schema = "zb;auth" }.Validate());
+
+    [Fact]
+    public void Schema_DefaultsToZbauth() =>
+        Assert.Equal("zbauth", new SqlServerApiKeyOptions().Schema);
+}
@@ -0,0 +1,66 @@
+using Microsoft.Extensions.Configuration;
+using Microsoft.Extensions.DependencyInjection;
+using Microsoft.Extensions.Hosting;
+using Microsoft.Extensions.Options;
+using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
+using ZB.MOM.WW.Auth.ApiKeys.SqlServer.DependencyInjection;
+
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests;
+
+public sealed class SqlServerApiKeyServiceCollectionExtensionsTests
+{
+    private static IConfiguration MinimalConfig() =>
+        new ConfigurationBuilder().AddInMemoryCollection(new Dictionary<string, string?>
+        {
+            ["ApiKeys:TokenPrefix"] = "sbk",
+            ["ApiKeys:PepperSecretName"] = "Pepper",
+            ["Pepper"] = "0123456789abcdef",
+            ["ApiKeys:SqlServer:ConnectionString"] = "Server=unused;TrustServerCertificate=true",
+        }).Build();
+
+    [Fact]
+    public void AddZbApiKeyAuthSqlServer_DisplacesSqliteStoresAndMigrator()
+    {
+        var services = new ServiceCollection();
+        services.AddZbApiKeyAuthSqlServer(MinimalConfig(), "ApiKeys", "ApiKeys:SqlServer");
+        using ServiceProvider sp = services.BuildServiceProvider();
+
+        Assert.IsType<SqlServerApiKeyStore>(sp.GetRequiredService<IApiKeyStore>());
+        Assert.IsType<SqlServerApiKeyAdminStore>(sp.GetRequiredService<IApiKeyAdminStore>());
+        Assert.IsType<SqlServerApiKeyAuditStore>(sp.GetRequiredService<IApiKeyAuditStore>());
+        Assert.IsType<SqlServerApiKeyStoreMigrator>(sp.GetRequiredService<IApiKeyStoreMigrator>());
+        Assert.NotNull(sp.GetRequiredService<SqliteApiKeyStoreImporter>());
+        Assert.NotNull(sp.GetRequiredService<IApiKeyVerifier>());
+    }
+
+    [Fact]
+    public void AddZbApiKeyAuthSqlServer_BindsSqlServerOptionsFromSection()
+    {
+        var config = new ConfigurationBuilder().AddInMemoryCollection(new Dictionary<string, string?>
+        {
+            ["ApiKeys:TokenPrefix"] = "sbk",
+            ["ApiKeys:PepperSecretName"] = "Pepper",
+            ["Pepper"] = "0123456789abcdef",
+            ["ApiKeys:SqlServer:ConnectionString"] = "Server=my-shared-sql;TrustServerCertificate=true",
+            ["ApiKeys:SqlServer:Schema"] = "zbauth_custom",
+        }).Build();
+        var services = new ServiceCollection();
+        services.AddZbApiKeyAuthSqlServer(config, "ApiKeys", "ApiKeys:SqlServer");
+        using ServiceProvider sp = services.BuildServiceProvider();
+
+        SqlServerApiKeyOptions options = sp.GetRequiredService<IOptions<SqlServerApiKeyOptions>>().Value;
+        Assert.Equal("Server=my-shared-sql;TrustServerCertificate=true", options.ConnectionString);
+        Assert.Equal("zbauth_custom", options.Schema);
+    }
+
+    [Fact]
+    public void AddZbApiKeyAuthSqlServer_RegistersMigrationHostedService()
+    {
+        var services = new ServiceCollection();
+        services.AddZbApiKeyAuthSqlServer(MinimalConfig(), "ApiKeys", "ApiKeys:SqlServer");
+        using ServiceProvider sp = services.BuildServiceProvider();
+
+        IEnumerable<IHostedService> hostedServices = sp.GetServices<IHostedService>();
+        Assert.Contains(hostedServices, s => s.GetType().Name == "ApiKeyMigrationHostedService");
+    }
+}
@@ -0,0 +1,43 @@
+namespace ZB.MOM.WW.Auth.ApiKeys.SqlServer.Tests;
+
+/// <summary>
+/// Offline pins for the migrator's retry classification. A <see cref="Microsoft.Data.SqlClient.SqlException"/>
+/// cannot be constructed normally, so the error-number predicate is extracted as an internal static
+/// and pinned here by number; the concurrent-provision behaviour itself is exercised by the
+/// env-gated live suite. These run on any machine, so the retry filter stays pinned even when the
+/// live suite skips.
+/// </summary>
+public sealed class SqlServerApiKeyStoreMigratorTests
+{
+    /// <summary>
+    /// The provisioning-race outcomes the ScadaBridge central shared-store live gate (2026-08-07)
+    /// proved reachable — or that the multi-statement DDL makes reachable — when N nodes provision
+    /// one virgin database at once. Losing such a race means the object now exists, which is the
+    /// desired end state, so a retry of the idempotent IF NOT EXISTS path must be taken.
+    /// </summary>
+    /// <param name="errorNumber">The SQL Server error number under test.</param>
+    [Theory]
+    [InlineData(1205)] // deadlock victim
+    [InlineData(2714)] // "There is already an object named ..." — lost CREATE SCHEMA/TABLE (the live-gate defect)
+    [InlineData(1913)] // "index ... already exists" — lost CREATE INDEX
+    [InlineData(2627)] // PK violation — lost schema_version MERGE-insert
+    public void Provisioning_race_error_numbers_are_retryable(int errorNumber)
+        => Assert.True(SqlServerApiKeyStoreMigrator.IsRetryableMigrationErrorNumber(errorNumber));
+
+    /// <summary>
+    /// Errors that reproduce identically on retry must propagate so the node fails closed instead
+    /// of spinning through the retry budget against a genuine misconfiguration.
+    /// </summary>
+    /// <param name="errorNumber">The SQL Server error number under test.</param>
+    [Theory]
+    [InlineData(2705)] // duplicate column — unreachable from the race (single CREATE TABLE statement)
+    [InlineData(208)] // invalid object name
+    [InlineData(229)] // permission denied
+    [InlineData(2759)] // "CREATE SCHEMA failed due to previous errors" — retryable only via its companion 2714
+    [InlineData(4060)] // cannot open database
+    [InlineData(18456)] // login failed
+    [InlineData(0)]
+    [InlineData(-2)] // client-side timeout
+    public void Non_race_error_numbers_are_not_retryable(int errorNumber)
+        => Assert.False(SqlServerApiKeyStoreMigrator.IsRetryableMigrationErrorNumber(errorNumber));
+}
@@ -0,0 +1,30 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+  <PropertyGroup>
+    <IsPackable>false</IsPackable>
+  </PropertyGroup>
+
+  <ItemGroup>
+    <PackageReference Include="coverlet.collector" />
+    <PackageReference Include="Microsoft.NET.Test.Sdk" />
+    <PackageReference Include="xunit" />
+    <PackageReference Include="xunit.runner.visualstudio" />
+    <!-- Live tests (Task 3+) skip rather than fail when AUTH_SQLSERVER_CONNSTR is unset. -->
+    <PackageReference Include="Xunit.SkippableFact" />
+    <PackageReference Include="Microsoft.Extensions.DependencyInjection" />
+    <PackageReference Include="Microsoft.Extensions.Configuration" />
+    <PackageReference Include="Microsoft.Extensions.Configuration.Abstractions" />
+    <PackageReference Include="Microsoft.Extensions.Hosting.Abstractions" />
+  </ItemGroup>
+
+  <ItemGroup>
+    <Using Include="Xunit" />
+  </ItemGroup>
+
+  <ItemGroup>
+    <ProjectReference Include="..\..\src\ZB.MOM.WW.Auth.ApiKeys.SqlServer\ZB.MOM.WW.Auth.ApiKeys.SqlServer.csproj" />
+    <ProjectReference Include="..\..\src\ZB.MOM.WW.Auth.ApiKeys\ZB.MOM.WW.Auth.ApiKeys.csproj" />
+    <ProjectReference Include="..\..\src\ZB.MOM.WW.Auth.Abstractions\ZB.MOM.WW.Auth.Abstractions.csproj" />
+  </ItemGroup>
+
+</Project>
@@ -5,6 +5,7 @@ using Microsoft.Extensions.Hosting;
 using Microsoft.Extensions.Options;
 using ZB.MOM.WW.Auth.Abstractions.ApiKeys;
 using ZB.MOM.WW.Auth.ApiKeys.DependencyInjection;
+using ZB.MOM.WW.Auth.ApiKeys.Sqlite;

 namespace ZB.MOM.WW.Auth.ApiKeys.Tests;

@@ -197,4 +198,37 @@ public class ApiKeyServiceCollectionExtensionsTests
                File.Delete(sqlitePath);
        }
    }
+
+    [Fact]
+    public void AddZbApiKeyAuth_RegistersSqliteMigratorAsIApiKeyStoreMigrator()
+    {
+        IConfiguration config = BuildConfiguration(TempSqlitePath());
+        var services = new ServiceCollection();
+
+        services.AddZbApiKeyAuth(config, ApiKeySection);
+
+        using ServiceProvider sp = services.BuildServiceProvider();
+
+        Assert.IsType<SqliteAuthStoreMigrator>(sp.GetRequiredService<IApiKeyStoreMigrator>());
+        Assert.Same(sp.GetRequiredService<SqliteAuthStoreMigrator>(), sp.GetRequiredService<IApiKeyStoreMigrator>());
+    }
+
+    [Fact]
+    public void AddZbApiKeyAuth_LeavesPreRegisteredMigratorInPlace()
+    {
+        var services = new ServiceCollection();
+        var fake = new FakeMigrator();
+        services.AddSingleton<IApiKeyStoreMigrator>(fake);
+
+        services.AddZbApiKeyAuth(BuildConfiguration(TempSqlitePath()), ApiKeySection);
+
+        using ServiceProvider sp = services.BuildServiceProvider();
+
+        Assert.Same(fake, sp.GetRequiredService<IApiKeyStoreMigrator>());
+    }
+
+    private sealed class FakeMigrator : IApiKeyStoreMigrator
+    {
+        public Task MigrateAsync(CancellationToken ct) => Task.CompletedTask;
+    }
 }
@@ -134,4 +134,30 @@ public class ServiceCollectionExtensionsTests
        await host.StartAsync();
        await host.StopAsync();
    }
+
+    [Fact]
+    public void AddZbLdapAuth_BindsUserBindModeKeys()
+    {
+        IConfiguration config = new ConfigurationBuilder()
+            .AddInMemoryCollection(new Dictionary<string, string?>
+            {
+                [$"{LdapSection}:Server"] = LdapServer,
+                [$"{LdapSection}:SearchBase"] = "dc=example,dc=com",
+                [$"{LdapSection}:Transport"] = nameof(LdapTransport.Ldaps),
+                [$"{LdapSection}:BindMode"] = "User",
+                [$"{LdapSection}:UserBindTemplate"] = "{0}@corp.test",
+                [$"{LdapSection}:AcceptDomainQualifiedNames"] = "false",
+            })
+            .Build();
+        var services = new ServiceCollection();
+
+        services.AddZbLdapAuth(config, LdapSection);
+
+        using ServiceProvider provider = services.BuildServiceProvider();
+        var options = provider.GetRequiredService<IOptions<LdapOptions>>().Value;
+
+        Assert.Equal(LdapBindMode.User, options.BindMode);
+        Assert.Equal("{0}@corp.test", options.UserBindTemplate);
+        Assert.False(options.AcceptDomainQualifiedNames);
+    }
 }
@@ -16,6 +16,8 @@ internal sealed class FakeLdapConnection : ILdapConnection
    private bool _throwOnConnect;
    private bool _throwOnServiceBind;
    private bool _throwOnUserBind;
+    private int? _firstBindResultCode;
+    private bool _throwOnSearch;

    // ---- observation -----

@@ -32,13 +34,24 @@ internal sealed class FakeLdapConnection : ILdapConnection
    /// </summary>
    public int BindAttempts { get; private set; }

+    /// <summary>The attribute list passed to the most recent <see cref="Search"/> call (the fake still returns every scripted attribute).</summary>
+    public IReadOnlyList<string>? SearchAttributes { get; private set; }
+
+    /// <summary>Every call in order: <c>connect</c>, <c>bind:&lt;dn&gt;</c>, <c>search:&lt;filter&gt;</c>.</summary>
+    public List<string> Calls { get; } = new();
+
    // ---- builder methods -----

    /// <summary>
    /// Scripts a user entry that will be returned by the next <see cref="Search"/> call.
-    /// Builds a minimal attribute bag with <c>memberOf</c> and optional <c>displayName</c>.
+    /// Builds a minimal attribute bag with <c>memberOf</c>, optional <c>displayName</c> and any
+    /// <paramref name="extraAttributes"/> (e.g. <c>sAMAccountName</c>) the test needs.
    /// </summary>
-    public FakeLdapConnection WithUserEntry(string dn, string[] memberOf, string? displayName = null)
+    public FakeLdapConnection WithUserEntry(
+        string dn,
+        string[] memberOf,
+        string? displayName = null,
+        IReadOnlyDictionary<string, string[]>? extraAttributes = null)
    {
        var attrs = new Dictionary<string, IReadOnlyList<string>>(StringComparer.OrdinalIgnoreCase)
        {
@@ -46,6 +59,11 @@ internal sealed class FakeLdapConnection : ILdapConnection
        };
        if (displayName is not null)
            attrs["displayName"] = new[] { displayName };
+        if (extraAttributes is not null)
+        {
+            foreach (var (name, values) in extraAttributes)
+                attrs[name] = values;
+        }

        _scriptedEntries.Add(new LdapSearchEntry(dn, attrs));
        return this;
@@ -83,6 +101,24 @@ internal sealed class FakeLdapConnection : ILdapConnection
        return this;
    }

+    /// <summary>
+    /// Throw <see cref="Novell.Directory.Ldap.LdapException"/> with <paramref name="resultCode"/> on the
+    /// FIRST bind. In user-bind mode the first bind is the user's own, so 49 simulates a wrong
+    /// password and e.g. 52 (Unavailable) a domain controller that cannot answer.
+    /// </summary>
+    public FakeLdapConnection ThrowOnFirstBind(int resultCode)
+    {
+        _firstBindResultCode = resultCode;
+        return this;
+    }
+
+    /// <summary>Throw <see cref="Novell.Directory.Ldap.LdapException"/> from <see cref="Search"/> (unhealthy directory).</summary>
+    public FakeLdapConnection ThrowOnSearch()
+    {
+        _throwOnSearch = true;
+        return this;
+    }
+
    /// <summary>
    /// Throw <see cref="Novell.Directory.Ldap.LdapException"/> from <see cref="Connect"/> to
    /// simulate an unreachable directory (infrastructure failure).
@@ -120,6 +156,7 @@ internal sealed class FakeLdapConnection : ILdapConnection
        int timeoutMs,
        RemoteCertificateValidationCallback? serverCertificateValidationCallback = null)
    {
+        Calls.Add("connect");
        ConnectArgs = (host, port, transport, allowInsecure, timeoutMs);
        ConnectCertCallback = serverCertificateValidationCallback;
        if (_throwOnConnect)
@@ -130,8 +167,12 @@ internal sealed class FakeLdapConnection : ILdapConnection
    public void Bind(string dn, string password)
    {
        BindAttempts++;
+        Calls.Add($"bind:{dn}");
        var isServiceBind = BindAttempts == 1;

+        if (isServiceBind && _firstBindResultCode is { } code)
+            throw new Novell.Directory.Ldap.LdapException("Bind refused", code, dn);
+
        if ((_throwOnServiceBind && isServiceBind)
            || (_throwOnUserBind && !isServiceBind)
            || _throwBindDns.Contains(dn))
@@ -147,7 +188,14 @@ internal sealed class FakeLdapConnection : ILdapConnection
        string searchBase,
        string filter,
        IReadOnlyList<string> attributes)
-        => _scriptedEntries.AsReadOnly();
+    {
+        Calls.Add($"search:{filter}");
+        SearchAttributes = attributes;
+        if (_throwOnSearch)
+            throw new Novell.Directory.Ldap.LdapException(
+                "Search failed", Novell.Directory.Ldap.LdapException.Unavailable, searchBase);
+        return _scriptedEntries.AsReadOnly();
+    }

    public void Dispose() { /* nothing to clean up */ }
 }
@@ -99,4 +99,70 @@ public sealed class GLAuthIntegrationTests
            $"Authentication failed: {result.Failure} (server={server}:{port}, user={user})");
        Assert.NotEmpty(result.Groups);
    }
+
+    /// <summary>
+    /// Failover: primary endpoint points at a closed local port (refused immediately),
+    /// fallback at the real GLAuth. The login must succeed via the fallback, and a second
+    /// login must also succeed (sticky preference — no re-walk through the dead primary).
+    /// Skipped unless <c>ZB_LDAP_IT=1</c> is set; skipped again if the server is unreachable.
+    /// </summary>
+    [SkippableFact]
+    public async Task Authenticate_FailsOverFromDeadPrimary_ToRealGLAuth()
+    {
+        // ------------------------------------------------------------------ opt-in gate
+        Skip.IfNot(
+            Environment.GetEnvironmentVariable("ZB_LDAP_IT") == "1",
+            "Set ZB_LDAP_IT=1 and a reachable GLAuth to run.");
+
+        // ------------------------------------------------------------------ read config
+        var server   = Environment.GetEnvironmentVariable("ZB_LDAP_SERVER")   ?? "localhost";
+        var port     = int.TryParse(Environment.GetEnvironmentVariable("ZB_LDAP_PORT"), out var p) ? p : 3893;
+        var baseDn   = Environment.GetEnvironmentVariable("ZB_LDAP_BASE")     ?? "dc=lmxopcua,dc=local";
+        var svcDn    = Environment.GetEnvironmentVariable("ZB_LDAP_SVC_DN")   ?? "";
+        var svcPw    = Environment.GetEnvironmentVariable("ZB_LDAP_SVC_PW")   ?? "";
+        var user     = Environment.GetEnvironmentVariable("ZB_LDAP_USER")     ?? "";
+        var pw       = Environment.GetEnvironmentVariable("ZB_LDAP_PW")       ?? "";
+        var userAttr = Environment.GetEnvironmentVariable("ZB_LDAP_USERATTR") ?? "cn";
+
+        // ------------------------------------------------------------------ reachability probe
+        try
+        {
+            using var tcp = new TcpClient();
+            // 3-second connect timeout to keep the test suite snappy when the server is absent
+            var connectTask = tcp.ConnectAsync(server, port);
+            if (!connectTask.Wait(TimeSpan.FromSeconds(3)))
+                Skip.If(true, $"GLAuth not reachable at {server}:{port} (connect timed out).");
+        }
+        catch (Exception ex)
+        {
+            Skip.If(true, $"GLAuth not reachable at {server}:{port}: {ex.Message}");
+        }
+
+        // ------------------------------------------------------------------ build options
+        var options = new LdapOptions
+        {
+            Enabled                = true,
+            Server                 = "127.0.0.1",
+            Port                   = 39999,          // closed port -> immediate refusal
+            Transport              = LdapTransport.None,
+            AllowInsecure          = true,
+            SearchBase             = baseDn,
+            ServiceAccountDn       = svcDn,
+            ServiceAccountPassword = svcPw,
+            UserNameAttribute      = userAttr,
+            GroupAttribute         = "memberOf",
+            ConnectionTimeoutMs    = 3000,
+            FallbackServers        = new[] { $"{server}:{port}" },
+        };
+
+        // ------------------------------------------------------------------ exercise the real service
+        var svc = new LdapAuthService(options);
+
+        var first = await svc.AuthenticateAsync(user, pw, CancellationToken.None);
+        Assert.True(first.Succeeded, $"first login failed: {first.Failure}");
+
+        // Second login exercises the sticky preference: it must not re-walk the dead primary.
+        var second = await svc.AuthenticateAsync(user, pw, CancellationToken.None);
+        Assert.True(second.Succeeded, $"second login failed: {second.Failure}");
+    }
 }
@@ -0,0 +1,297 @@
+using ZB.MOM.WW.Auth.Abstractions.Ldap;
+
+namespace ZB.MOM.WW.Auth.Ldap.Tests;
+
+/// <summary>
+/// Covers the multi-endpoint failover walk: which failures fail over (system-side only),
+/// which are final (any authoritative answer from a healthy directory), how the sticky
+/// endpoint preference moves, and that single-endpoint behaviour is unchanged.
+/// </summary>
+public class LdapAuthServiceFailoverTests
+{
+    private static LdapOptions Opts(params string[] fallbacks) => new()
+    {
+        Enabled = true,
+        Server = "dc1",
+        Port = 636,
+        Transport = LdapTransport.Ldaps,
+        SearchBase = "dc=x",
+        ServiceAccountDn = "cn=svc,dc=x",
+        ServiceAccountPassword = "svcpw",
+        FallbackServers = fallbacks,
+    };
+
+    private static FakeLdapConnection HealthyWithUser() =>
+        new FakeLdapConnection().WithUserEntry("cn=alice,dc=x", new[] { "cn=ops,dc=x" });
+
+    // ---- failover triggers (system-side) ----
+
+    [Fact]
+    public async Task ConnectFailure_FailsOverToFallback_AndSucceeds()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnConnect(), HealthyWithUser());
+        var svc = new LdapAuthService(Opts("dc2:3894"), factory);
+
+        var result = await svc.AuthenticateAsync("alice", "pw", CancellationToken.None);
+
+        Assert.True(result.Succeeded);
+        Assert.Equal(2, factory.Created.Count);
+        Assert.Equal("dc1", factory.Created[0].ConnectArgs!.Value.Host);
+        Assert.Equal("dc2", factory.Created[1].ConnectArgs!.Value.Host);
+        Assert.Equal(3894, factory.Created[1].ConnectArgs!.Value.Port);
+    }
+
+    [Fact]
+    public async Task ServiceBindFailure_FailsOverToFallback()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnServiceBind(), HealthyWithUser());
+        var svc = new LdapAuthService(Opts("dc2"), factory);
+
+        var result = await svc.AuthenticateAsync("alice", "pw", CancellationToken.None);
+
+        Assert.True(result.Succeeded);
+        Assert.Equal(636, factory.Created[1].ConnectArgs!.Value.Port); // bare host adopts Port
+    }
+
+    // ---- user-side results NEVER fail over ----
+
+    [Fact]
+    public async Task BadUserCredentials_DoesNotFailOver()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            HealthyWithUser().ThrowOnUserBind());
+        var svc = new LdapAuthService(Opts("dc2"), factory);
+
+        var result = await svc.AuthenticateAsync("alice", "wrong", CancellationToken.None);
+
+        Assert.Equal(LdapAuthFailure.BadCredentials, result.Failure);
+        Assert.Single(factory.Created);
+    }
+
+    [Fact]
+    public async Task UserNotFound_DoesNotFailOver()
+    {
+        var factory = new SequenceLdapConnectionFactory(new FakeLdapConnection());
+        var svc = new LdapAuthService(Opts("dc2"), factory);
+
+        var result = await svc.AuthenticateAsync("ghost", "pw", CancellationToken.None);
+
+        Assert.Equal(LdapAuthFailure.UserNotFound, result.Failure);
+        Assert.Single(factory.Created);
+    }
+
+    [Fact]
+    public async Task GroupLookupFailed_DoesNotFailOver()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().WithUserEntry("cn=alice,dc=x", Array.Empty<string>()));
+        var svc = new LdapAuthService(Opts("dc2"), factory);
+
+        var result = await svc.AuthenticateAsync("alice", "pw", CancellationToken.None);
+
+        Assert.Equal(LdapAuthFailure.GroupLookupFailed, result.Failure);
+        Assert.Single(factory.Created);
+    }
+
+    [Fact]
+    public async Task AmbiguousUser_DoesNotFailOver()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().WithDuplicateMatch());
+        var svc = new LdapAuthService(Opts("dc2"), factory);
+
+        var result = await svc.AuthenticateAsync("alice", "pw", CancellationToken.None);
+
+        Assert.Equal(LdapAuthFailure.AmbiguousUser, result.Failure);
+        Assert.Single(factory.Created);
+    }
+
+    // ---- exhaustion ----
+
+    [Fact]
+    public async Task AllEndpointsDown_ReturnsServiceAccountBindFailed()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnConnect(),
+            new FakeLdapConnection().ThrowOnConnect());
+        var svc = new LdapAuthService(Opts("dc2"), factory);
+
+        var result = await svc.AuthenticateAsync("alice", "pw", CancellationToken.None);
+
+        Assert.Equal(LdapAuthFailure.ServiceAccountBindFailed, result.Failure);
+        Assert.Equal(2, factory.Created.Count);
+    }
+
+    // ---- stickiness ----
+
+    [Fact]
+    public async Task StickyPreference_SecondLoginStartsAtLastGoodEndpoint()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnConnect(), // login 1 @ dc1: dead
+            HealthyWithUser(),                          // login 1 @ dc2: answers
+            HealthyWithUser());                         // login 2: should go straight to dc2
+        var svc = new LdapAuthService(Opts("dc2"), factory);
+
+        Assert.True((await svc.AuthenticateAsync("alice", "pw", CancellationToken.None)).Succeeded);
+        Assert.True((await svc.AuthenticateAsync("alice", "pw", CancellationToken.None)).Succeeded);
+
+        Assert.Equal(3, factory.Created.Count);
+        Assert.Equal("dc2", factory.Created[2].ConnectArgs!.Value.Host);
+    }
+
+    [Fact]
+    public async Task StickyPreference_DriftsBackToPrimary_WhenPreferredFails()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnConnect(), // login 1 @ dc1: dead
+            HealthyWithUser(),                          // login 1 @ dc2: answers -> preferred
+            new FakeLdapConnection().ThrowOnConnect(), // login 2 @ dc2: now dead
+            HealthyWithUser(),                          // login 2 @ dc1 (declared order): recovered
+            HealthyWithUser());                         // login 3: should start at dc1 again
+        var svc = new LdapAuthService(Opts("dc2"), factory);
+
+        Assert.True((await svc.AuthenticateAsync("alice", "pw", CancellationToken.None)).Succeeded);
+        Assert.True((await svc.AuthenticateAsync("alice", "pw", CancellationToken.None)).Succeeded);
+        Assert.True((await svc.AuthenticateAsync("alice", "pw", CancellationToken.None)).Succeeded);
+
+        Assert.Equal(5, factory.Created.Count);
+        Assert.Equal("dc1", factory.Created[3].ConnectArgs!.Value.Host);
+        Assert.Equal("dc1", factory.Created[4].ConnectArgs!.Value.Host);
+    }
+
+    [Fact]
+    public async Task StickyPreference_UpdatesOnUserSideAnswer_NotOnlySuccess()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnConnect(), // login 1 @ dc1: dead
+            new FakeLdapConnection(),                   // login 1 @ dc2: answers UserNotFound
+            HealthyWithUser());                         // login 2: should start at dc2
+        var svc = new LdapAuthService(Opts("dc2"), factory);
+
+        var first = await svc.AuthenticateAsync("ghost", "pw", CancellationToken.None);
+        Assert.Equal(LdapAuthFailure.UserNotFound, first.Failure);
+
+        Assert.True((await svc.AuthenticateAsync("alice", "pw", CancellationToken.None)).Succeeded);
+        Assert.Equal("dc2", factory.Created[2].ConnectArgs!.Value.Host);
+    }
+
+    [Fact]
+    public async Task AllEndpointsDown_LeavesPreferenceUnchanged()
+    {
+        // A total outage teaches nothing about which endpoint is healthy, so the walk
+        // must not move the preference — the next login still starts at the primary.
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnConnect(), // login 1 @ dc1
+            new FakeLdapConnection().ThrowOnConnect(), // login 1 @ dc2
+            HealthyWithUser());                         // login 2: must start at dc1 again
+        var svc = new LdapAuthService(Opts("dc2"), factory);
+
+        Assert.Equal(LdapAuthFailure.ServiceAccountBindFailed,
+            (await svc.AuthenticateAsync("alice", "pw", CancellationToken.None)).Failure);
+        Assert.True((await svc.AuthenticateAsync("alice", "pw", CancellationToken.None)).Succeeded);
+
+        Assert.Equal("dc1", factory.Created[2].ConnectArgs!.Value.Host);
+    }
+
+    // ---- back-compat ----
+
+    [Fact]
+    public async Task NoFallbacks_SingleAttempt_UnchangedBehavior()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnConnect());
+        var svc = new LdapAuthService(Opts(/* no fallbacks */), factory);
+
+        var result = await svc.AuthenticateAsync("alice", "pw", CancellationToken.None);
+
+        Assert.Equal(LdapAuthFailure.ServiceAccountBindFailed, result.Failure);
+        Assert.Single(factory.Created);
+    }
+
+    [Fact]
+    public async Task Disabled_NeverOpensAConnection_EvenWithFallbacks()
+    {
+        var factory = new SequenceLdapConnectionFactory(HealthyWithUser());
+        var svc = new LdapAuthService(Opts("dc2") with { Enabled = false }, factory);
+
+        var result = await svc.AuthenticateAsync("alice", "pw", CancellationToken.None);
+
+        Assert.Equal(LdapAuthFailure.Disabled, result.Failure);
+        Assert.Empty(factory.Created);
+    }
+
+    [Fact]
+    public async Task MalformedFallbackEntry_IsSkippedDefensively()
+    {
+        // The validator rejects this at boot; a service constructed without validation
+        // must degrade to the parseable subset, not throw.
+        var factory = new SequenceLdapConnectionFactory(HealthyWithUser());
+        var svc = new LdapAuthService(Opts(":::"), factory);
+
+        Assert.True((await svc.AuthenticateAsync("alice", "pw", CancellationToken.None)).Succeeded);
+        Assert.Single(factory.Created);
+    }
+
+    // ---- observability ----
+
+    /// <summary>Captures emitted log entries so failover visibility can be asserted.</summary>
+    private sealed class RecordingLogger : Microsoft.Extensions.Logging.ILogger<LdapAuthService>
+    {
+        public List<(Microsoft.Extensions.Logging.LogLevel Level, string Message)> Entries { get; } = new();
+        public IDisposable? BeginScope<TState>(TState state) where TState : notnull => null;
+        public bool IsEnabled(Microsoft.Extensions.Logging.LogLevel logLevel) => true;
+        public void Log<TState>(Microsoft.Extensions.Logging.LogLevel logLevel, Microsoft.Extensions.Logging.EventId eventId,
+            TState state, Exception? exception, Func<TState, Exception?, string> formatter)
+            => Entries.Add((logLevel, formatter(state, exception)));
+    }
+
+    [Fact]
+    public async Task Failover_LogsWarningWithEndpointAndStage_AndInfoOnStickyChange()
+    {
+        var logger = new RecordingLogger();
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnConnect(), HealthyWithUser());
+        var svc = new LdapAuthService(Opts("dc2:3894"), factory, logger);
+
+        await svc.AuthenticateAsync("alice", "pw", CancellationToken.None);
+
+        Assert.Contains(logger.Entries, e =>
+            e.Level == Microsoft.Extensions.Logging.LogLevel.Warning
+            && e.Message.Contains("dc1:636") && e.Message.Contains("connect"));
+        Assert.Contains(logger.Entries, e =>
+            e.Level == Microsoft.Extensions.Logging.LogLevel.Information
+            && e.Message.Contains("dc2:3894"));
+        // Never log secrets:
+        Assert.DoesNotContain(logger.Entries, e => e.Message.Contains("svcpw") || e.Message.Contains("pw\""));
+    }
+
+    [Fact]
+    public async Task HappyPath_LogsNothing()
+    {
+        var logger = new RecordingLogger();
+        var factory = new SequenceLdapConnectionFactory(HealthyWithUser());
+        var svc = new LdapAuthService(Opts("dc2"), factory, logger);
+
+        await svc.AuthenticateAsync("alice", "pw", CancellationToken.None);
+
+        Assert.Empty(logger.Entries);
+    }
+
+    [Fact]
+    public async Task ServiceBindFailure_LogsTheServiceBindStage_NotConnect()
+    {
+        var logger = new RecordingLogger();
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnServiceBind(), HealthyWithUser());
+        var svc = new LdapAuthService(Opts("dc2"), factory, logger);
+
+        await svc.AuthenticateAsync("alice", "pw", CancellationToken.None);
+
+        Assert.Contains(logger.Entries, e =>
+            e.Level == Microsoft.Extensions.Logging.LogLevel.Warning
+            && e.Message.Contains("dc1:636") && e.Message.Contains("service-bind"));
+    }
+}
@@ -147,4 +147,45 @@ public class LdapAuthServiceTests
        Assert.True(r.Succeeded);
        Assert.Equal(new[] { "Eng,ineers", "A,B" }, r.Groups);
    }
+
+    [Fact]
+    public async Task Username_IsCanonicalisedFromTheEntry_InServiceAccountMode()
+    {
+        var fake = new FakeLdapConnection().WithUserEntry(
+            "cn=alice,dc=x", new[] { "cn=g,dc=x" }, "Alice",
+            new Dictionary<string, string[]> { ["cn"] = new[] { "alice" } });
+        var svc = new LdapAuthService(Opts(), new FakeLdapConnectionFactory(fake));
+
+        var r = await svc.AuthenticateAsync("ALICE", "pw", default);
+
+        Assert.True(r.Succeeded);
+        Assert.Equal("alice", r.Username);
+    }
+
+    [Fact]
+    public async Task Username_FallsBackToTheTypedName_InServiceAccountMode_WhenAttributeAbsent()
+    {
+        var fake = new FakeLdapConnection().WithUserEntry("cn=alice,dc=x", new[] { "cn=g,dc=x" }, "Alice");
+        var svc = new LdapAuthService(Opts(), new FakeLdapConnectionFactory(fake));
+
+        var r = await svc.AuthenticateAsync("ALICE", "pw", default);
+
+        Assert.True(r.Succeeded);
+        Assert.Equal("ALICE", r.Username);
+    }
+
+    [Theory]
+    [InlineData("")]
+    [InlineData(null)]
+    public async Task EmptyPassword_IsRefused_InServiceAccountMode_WithoutContactingTheDirectory(string? password)
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().WithUserEntry("cn=alice,dc=x", new[] { "cn=g,dc=x" }, "Alice"));
+        var svc = new LdapAuthService(Opts(), factory);
+
+        var r = await svc.AuthenticateAsync("alice", password!, default);
+
+        Assert.Equal(LdapAuthFailure.BadCredentials, r.Failure);
+        Assert.Empty(factory.Created);
+    }
 }
@@ -0,0 +1,222 @@
+using Novell.Directory.Ldap;
+using ZB.MOM.WW.Auth.Abstractions.Ldap;
+
+namespace ZB.MOM.WW.Auth.Ldap.Tests;
+
+/// <summary>
+/// <see cref="LdapBindMode.User"/>: bind as the signing-in user, then read their own entry on the
+/// same connection. The service account in <see cref="Opts"/> is deliberately populated so the
+/// tests prove it is never used.
+/// </summary>
+public class LdapAuthServiceUserBindTests
+{
+    private static LdapOptions Opts(params string[] fallbacks) => new()
+    {
+        Enabled = true,
+        Server = "dc1",
+        Port = 636,
+        Transport = LdapTransport.Ldaps,
+        SearchBase = "dc=x",
+        UserNameAttribute = "sAMAccountName",
+        DisplayNameAttribute = "displayName",
+        GroupAttribute = "memberOf",
+        BindMode = LdapBindMode.User,
+        UserBindTemplate = "{0}@corp.test",
+        ServiceAccountDn = "cn=svc,dc=x",
+        ServiceAccountPassword = "svcpw",
+        FallbackServers = fallbacks,
+    };
+
+    private static FakeLdapConnection Jane() => new FakeLdapConnection()
+        .WithUserEntry("CN=Jane Doe,OU=u,DC=x", new[] { "CN=UG_A,OU=g,DC=x", "CN=UG_B,OU=g,DC=x" }, "Jane Doe");
+
+    private static Task<LdapAuthResult> Login(LdapAuthService svc, string user, string pw = "pw") =>
+        svc.AuthenticateAsync(user, pw, CancellationToken.None);
+
+    [Fact]
+    public async Task Success_BindsAsUser_ThenSearchesOnTheSameConnection()
+    {
+        var factory = new SequenceLdapConnectionFactory(Jane());
+        var result = await Login(new LdapAuthService(Opts(), factory), "jdoe");
+
+        Assert.True(result.Succeeded);
+        Assert.Equal("jdoe", result.Username);
+        Assert.Equal("Jane Doe", result.DisplayName);
+        Assert.Equal(new[] { "UG_A", "UG_B" }, result.Groups);
+        var conn = Assert.Single(factory.Created);
+        Assert.Equal(new[] { "connect", "bind:jdoe@corp.test", "search:(sAMAccountName=jdoe)" }, conn.Calls);
+        Assert.Equal(new[] { "jdoe@corp.test" }, conn.BoundDns);
+    }
+
+    [Theory]
+    [InlineData(@"NAM\jdoe")]
+    [InlineData("jdoe@corp.test")]
+    [InlineData("  jdoe ")]
+    public async Task QualifiedNames_BindAsTheTemplate_AndReturnTheBareName(string typed)
+    {
+        var factory = new SequenceLdapConnectionFactory(Jane());
+        var result = await Login(new LdapAuthService(Opts(), factory), typed);
+
+        Assert.True(result.Succeeded);
+        Assert.Equal("jdoe", result.Username);
+        Assert.Equal(new[] { "jdoe@corp.test" }, factory.Created[0].BoundDns);
+    }
+
+    [Fact]
+    public async Task EmptyPassword_IsRefused_WithoutContactingTheDirectory()
+    {
+        var factory = new SequenceLdapConnectionFactory(Jane());
+        var result = await Login(new LdapAuthService(Opts(), factory), "jdoe", "");
+
+        Assert.Equal(LdapAuthFailure.BadCredentials, result.Failure);
+        Assert.Empty(factory.Created);
+    }
+
+    [Theory]
+    [InlineData("@x")]
+    [InlineData(@"a\b\c")]
+    [InlineData("a@b@c")]
+    public async Task UnusableNames_AreRefused_WithoutContactingTheDirectory(string typed)
+    {
+        var factory = new SequenceLdapConnectionFactory(Jane());
+        var result = await Login(new LdapAuthService(Opts(), factory), typed);
+
+        Assert.Equal(LdapAuthFailure.BadCredentials, result.Failure);
+        Assert.Empty(factory.Created);
+    }
+
+    [Fact]
+    public async Task StrictMode_RefusesQualifiedNames_WithoutContactingTheDirectory()
+    {
+        var factory = new SequenceLdapConnectionFactory(Jane());
+        var svc = new LdapAuthService(Opts() with { AcceptDomainQualifiedNames = false }, factory);
+
+        var result = await Login(svc, @"NAM\jdoe");
+
+        Assert.Equal(LdapAuthFailure.BadCredentials, result.Failure);
+        Assert.Empty(factory.Created);
+    }
+
+    [Fact]
+    public async Task WrongPassword_IsFinal_AndNeverAsksTheFallback()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            Jane().ThrowOnFirstBind(LdapException.InvalidCredentials), Jane());
+        var result = await Login(new LdapAuthService(Opts("dc2"), factory), "jdoe", "wrong");
+
+        Assert.Equal(LdapAuthFailure.BadCredentials, result.Failure);
+        Assert.Single(factory.Created);
+    }
+
+    [Fact]
+    public async Task NonCredentialBindError_FailsOver()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            Jane().ThrowOnFirstBind(LdapException.Unavailable), Jane());
+        var result = await Login(new LdapAuthService(Opts("dc2"), factory), "jdoe");
+
+        Assert.True(result.Succeeded);
+        Assert.Equal(2, factory.Created.Count);
+        Assert.Equal("dc2", factory.Created[1].ConnectArgs!.Value.Host);
+    }
+
+    [Fact]
+    public async Task ConnectFailure_FailsOver()
+    {
+        var factory = new SequenceLdapConnectionFactory(new FakeLdapConnection().ThrowOnConnect(), Jane());
+        var result = await Login(new LdapAuthService(Opts("dc2"), factory), "jdoe");
+
+        Assert.True(result.Succeeded);
+        Assert.Equal(2, factory.Created.Count);
+    }
+
+    [Fact]
+    public async Task SearchFailure_FailsOver()
+    {
+        var factory = new SequenceLdapConnectionFactory(Jane().ThrowOnSearch(), Jane());
+        var result = await Login(new LdapAuthService(Opts("dc2"), factory), "jdoe");
+
+        Assert.True(result.Succeeded);
+        Assert.Equal(2, factory.Created.Count);
+    }
+
+    [Fact]
+    public async Task EveryEndpointDown_IsSystemSide()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().ThrowOnConnect(), new FakeLdapConnection().ThrowOnConnect());
+        var result = await Login(new LdapAuthService(Opts("dc2"), factory), "jdoe");
+
+        Assert.Equal(LdapAuthFailure.ServiceAccountBindFailed, result.Failure);
+    }
+
+    [Fact]
+    public async Task NoEntryAfterAGoodBind_IsGroupLookupFailed()
+    {
+        var factory = new SequenceLdapConnectionFactory(new FakeLdapConnection().WithNoMatch());
+        var result = await Login(new LdapAuthService(Opts("dc2"), factory), "jdoe");
+
+        Assert.Equal(LdapAuthFailure.GroupLookupFailed, result.Failure);
+        Assert.Single(factory.Created);
+    }
+
+    [Fact]
+    public async Task TwoEntriesAfterAGoodBind_IsGroupLookupFailed()
+    {
+        var factory = new SequenceLdapConnectionFactory(new FakeLdapConnection().WithDuplicateMatch());
+        var result = await Login(new LdapAuthService(Opts(), factory), "jdoe");
+
+        Assert.Equal(LdapAuthFailure.GroupLookupFailed, result.Failure);
+    }
+
+    [Fact]
+    public async Task NoGroups_IsGroupLookupFailed()
+    {
+        var factory = new SequenceLdapConnectionFactory(
+            new FakeLdapConnection().WithUserEntry("CN=Jane,DC=x", Array.Empty<string>(), "Jane"));
+        var result = await Login(new LdapAuthService(Opts(), factory), "jdoe");
+
+        Assert.Equal(LdapAuthFailure.GroupLookupFailed, result.Failure);
+    }
+
+    [Fact]
+    public async Task Username_IsCanonicalisedFromTheEntry_NotTheTypedCase()
+    {
+        var factory = new SequenceLdapConnectionFactory(new FakeLdapConnection().WithUserEntry(
+            "CN=Jane Doe,OU=u,DC=x", new[] { "CN=UG_A,OU=g,DC=x" }, "Jane Doe",
+            new Dictionary<string, string[]> { ["sAMAccountName"] = new[] { "jdoe" } }));
+        var result = await Login(new LdapAuthService(Opts(), factory), @"NAM\JDOE");
+
+        Assert.True(result.Succeeded);
+        Assert.Equal("jdoe", result.Username);
+    }
+
+    [Fact]
+    public async Task Username_FallsBackToTheTypedName_WhenTheEntryLacksTheAttribute()
+    {
+        var factory = new SequenceLdapConnectionFactory(Jane());
+        var result = await Login(new LdapAuthService(Opts(), factory), "JDOE");
+
+        Assert.True(result.Succeeded);
+        Assert.Equal("JDOE", result.Username);
+    }
+
+    [Fact]
+    public async Task Search_RequestsTheUserNameAttribute()
+    {
+        var factory = new SequenceLdapConnectionFactory(Jane());
+        await Login(new LdapAuthService(Opts(), factory), "jdoe");
+
+        Assert.Equal(new[] { "displayName", "memberOf", "sAMAccountName" }, factory.Created[0].SearchAttributes);
+    }
+
+    [Fact]
+    public async Task Search_DeDuplicatesTheUserNameAttribute_CaseInsensitively()
+    {
+        var factory = new SequenceLdapConnectionFactory(Jane());
+        var svc = new LdapAuthService(Opts() with { DisplayNameAttribute = "SAMACCOUNTNAME" }, factory);
+        await Login(svc, "jdoe");
+
+        Assert.Equal(new[] { "SAMACCOUNTNAME", "memberOf" }, factory.Created[0].SearchAttributes);
+    }
+}
@@ -0,0 +1,35 @@
+using ZB.MOM.WW.Auth.Ldap.Internal;
+
+namespace ZB.MOM.WW.Auth.Ldap.Tests;
+
+public class LdapEndpointParserTests
+{
+    [Theory]
+    [InlineData("dc2.example.com", 636, "dc2.example.com", 636)]      // bare host -> default port
+    [InlineData("dc2.example.com:3894", 636, "dc2.example.com", 3894)] // explicit port wins
+    [InlineData("  dc2.example.com : 3894  ", 636, "dc2.example.com", 3894)] // whitespace tolerated
+    public void TryParse_ValidEntry_Succeeds(string entry, int defaultPort, string host, int port)
+    {
+        Assert.True(LdapEndpointParser.TryParse(entry, defaultPort, out var ep, out _));
+        Assert.Equal(host, ep.Host);
+        Assert.Equal(port, ep.Port);
+    }
+
+    [Theory]
+    [InlineData("")]              // empty
+    [InlineData("   ")]           // whitespace
+    [InlineData(":636")]          // empty host
+    [InlineData("dc2:0")]         // port below range
+    [InlineData("dc2:65536")]     // port above range
+    [InlineData("dc2:abc")]       // non-numeric port
+    [InlineData("2001:db8::1:636")] // IPv6 literal (unsupported, >1 colon)
+    public void TryParse_InvalidEntry_Fails(string entry)
+    {
+        Assert.False(LdapEndpointParser.TryParse(entry, 636, out _, out var error));
+        Assert.NotEmpty(error);
+    }
+
+    [Fact]
+    public void Endpoint_ToString_IsHostColonPort() =>
+        Assert.Equal("dc2:3894", new LdapEndpoint("dc2", 3894).ToString());
+}
@@ -88,4 +88,66 @@ public class LdapOptionsValidatorTests
                ServiceAccountDn = "",
            })
            .Failed);
+
+    [Fact]
+    public void Validator_Succeeds_WithValidFallbackServers() =>
+        Assert.False(new LdapOptionsValidator()
+            .Validate(null, Opts() with { FallbackServers = new[] { "dc2", "dc3:3894" } })
+            .Failed);
+
+    [Theory]
+    [InlineData("")]
+    [InlineData(":636")]
+    [InlineData("dc2:0")]
+    [InlineData("dc2:notaport")]
+    public void Validator_Fails_OnMalformedFallbackServer(string entry)
+    {
+        var result = new LdapOptionsValidator()
+            .Validate(null, Opts() with { FallbackServers = new[] { entry } });
+
+        Assert.True(result.Failed);
+        Assert.Contains(nameof(LdapOptions.FallbackServers), result.FailureMessage);
+    }
+
+    [Fact]
+    public void Validator_Skips_FallbackServers_WhenDisabled() =>
+        // Same inert-when-disabled rule as every other connection field: a disabled
+        // provider never opens a connection, so its fallback list is never walked.
+        Assert.False(new LdapOptionsValidator()
+            .Validate(null, Opts() with { Enabled = false, FallbackServers = new[] { ":::" } })
+            .Failed);
+
+    private static LdapOptions UserOpts() => Opts() with
+    {
+        BindMode = LdapBindMode.User,
+        UserBindTemplate = "{0}@corp.test",
+        ServiceAccountDn = "",
+        ServiceAccountPassword = "",
+    };
+
+    [Fact]
+    public void Defaults_KeepServiceAccountMode()
+    {
+        var o = new LdapOptions();
+        Assert.Equal(LdapBindMode.ServiceAccount, o.BindMode);
+        Assert.Equal("", o.UserBindTemplate);
+        Assert.True(o.AcceptDomainQualifiedNames);
+    }
+
+    [Fact]
+    public void Validator_UserMode_DoesNotRequireServiceAccount() =>
+        Assert.True(new LdapOptionsValidator().Validate(null, UserOpts()).Succeeded);
+
+    [Theory]
+    [InlineData("")]
+    [InlineData("   ")]
+    [InlineData("jdoe@corp.test")]
+    [InlineData("{0}{0}@corp.test")]
+    public void Validator_UserMode_RequiresExactlyOnePlaceholder(string template)
+    {
+        var result = new LdapOptionsValidator().Validate(null, UserOpts() with { UserBindTemplate = template });
+
+        Assert.True(result.Failed);
+        Assert.Contains(nameof(LdapOptions.UserBindTemplate), result.FailureMessage, StringComparison.Ordinal);
+    }
 }
@@ -0,0 +1,47 @@
+using ZB.MOM.WW.Auth.Ldap.Internal;
+
+namespace ZB.MOM.WW.Auth.Ldap.Tests;
+
+public class LdapUserNameTests
+{
+    [Theory]
+    [InlineData("jdoe")]
+    [InlineData("  jdoe  ")]
+    [InlineData(@"NAM\jdoe")]
+    [InlineData("jdoe@corp.test")]
+    public void Accepting_ReducesToBareName(string input)
+    {
+        Assert.True(LdapUserName.TryNormalize(input, acceptDomainQualified: true, out var name));
+        Assert.Equal("jdoe", name);
+    }
+
+    [Theory]
+    [InlineData(@"NAM\jdoe")]
+    [InlineData("jdoe@corp.test")]
+    public void Strict_RefusesQualifiedNames(string input) =>
+        Assert.False(LdapUserName.TryNormalize(input, acceptDomainQualified: false, out _));
+
+    [Fact]
+    public void Strict_AcceptsBareName()
+    {
+        Assert.True(LdapUserName.TryNormalize(" jdoe ", acceptDomainQualified: false, out var name));
+        Assert.Equal("jdoe", name);
+    }
+
+    [Theory]
+    [InlineData(null)]
+    [InlineData("")]
+    [InlineData("   ")]
+    [InlineData("@x")]
+    [InlineData(@"\jdoe")]
+    [InlineData("jdoe@")]
+    [InlineData(@"NAM\")]
+    [InlineData(@"a\b\c")]
+    [InlineData("a@b@c")]
+    [InlineData(@"NAM\jdoe@corp.test")]
+    public void Unusable_IsRefused(string? input)
+    {
+        Assert.False(LdapUserName.TryNormalize(input, acceptDomainQualified: true, out var name));
+        Assert.Equal("", name);
+    }
+}
@@ -0,0 +1,179 @@
+using Novell.Directory.Ldap;
+using Xunit;
+using ZB.MOM.WW.Auth.Ldap.Internal;
+
+namespace ZB.MOM.WW.Auth.Ldap.Tests;
+
+/// <summary>
+/// Covers <see cref="NovellLdapConnection.ReadEntries"/>' handling of AD continuation
+/// referrals. Active Directory returns one for any search based at a domain root, and
+/// Novell surfaces it by throwing from <c>Next()</c>. Before this was handled, the
+/// exception escaped the search stage and every login against
+/// <c>DC=nam,DC=zimmer,DC=com</c> failed as <c>ServiceAccountBindFailed</c> while the
+/// directory and credentials were healthy.
+/// </summary>
+public sealed class NovellSearchReferralTests
+{
+    [Fact]
+    public void Referral_before_the_only_entry_does_not_lose_it()
+    {
+        var results = new ScriptedResults(
+            Step.Referral(),
+            Step.Entry("CN=dohertj2,OU=Office Users,OU=Warsaw,DC=nam,DC=zimmer,DC=com"));
+
+        var entries = NovellLdapConnection.ReadEntries(results);
+
+        var entry = Assert.Single(entries);
+        Assert.Equal("CN=dohertj2,OU=Office Users,OU=Warsaw,DC=nam,DC=zimmer,DC=com", entry.Dn);
+    }
+
+    [Fact]
+    public void Referral_after_the_entry_does_not_discard_what_was_already_read()
+    {
+        var results = new ScriptedResults(
+            Step.Entry("CN=a,DC=nam,DC=zimmer,DC=com"),
+            Step.Referral());
+
+        var entries = NovellLdapConnection.ReadEntries(results);
+
+        Assert.Equal("CN=a,DC=nam,DC=zimmer,DC=com", Assert.Single(entries).Dn);
+    }
+
+    /// <summary>
+    /// The real shape observed against the NAM domain root: several references interleaved
+    /// with the one user entry. All three of AD's usual continuation points are present.
+    /// </summary>
+    [Fact]
+    public void Multiple_interleaved_referrals_still_yield_every_entry()
+    {
+        var results = new ScriptedResults(
+            Step.Referral(),
+            Step.Entry("CN=a,DC=nam,DC=zimmer,DC=com"),
+            Step.Referral(),
+            Step.Referral(),
+            Step.Entry("CN=b,DC=nam,DC=zimmer,DC=com"));
+
+        var entries = NovellLdapConnection.ReadEntries(results);
+
+        Assert.Equal(["CN=a,DC=nam,DC=zimmer,DC=com", "CN=b,DC=nam,DC=zimmer,DC=com"],
+            entries.Select(e => e.Dn));
+    }
+
+    /// <summary>
+    /// A referral-only result set is a legitimate "nothing here" answer, and must come back
+    /// empty rather than throwing — <c>LdapAuthService</c> reads empty as <c>UserNotFound</c>,
+    /// which is the honest verdict, where an exception would become a system-side failure.
+    /// </summary>
+    [Fact]
+    public void Referrals_only_returns_empty_rather_than_throwing()
+    {
+        var results = new ScriptedResults(Step.Referral(), Step.Referral());
+
+        Assert.Empty(NovellLdapConnection.ReadEntries(results));
+    }
+
+    /// <summary>
+    /// A base <see cref="LdapException"/> carrying the Referral result code is treated the
+    /// same as the derived type — the catch filter must not depend on the exact class.
+    /// </summary>
+    [Fact]
+    public void Referral_result_code_on_a_base_exception_is_also_skipped()
+    {
+        var results = new ScriptedResults(
+            Step.Throwing(new LdapException("referral", LdapException.Referral, null)),
+            Step.Entry("CN=a,DC=nam,DC=zimmer,DC=com"));
+
+        Assert.Single(NovellLdapConnection.ReadEntries(results));
+    }
+
+    /// <summary>
+    /// Only referrals are swallowed. A genuine directory fault must still propagate so the
+    /// caller maps it to a system-side failure — silently returning a short result set would
+    /// turn "the directory is broken" into "no such user", which is the wrong answer to give
+    /// about someone's credentials.
+    /// </summary>
+    [Fact]
+    public void A_non_referral_fault_still_propagates()
+    {
+        var results = new ScriptedResults(
+            Step.Throwing(new LdapException("busy", LdapException.Busy, null)),
+            Step.Entry("CN=a,DC=nam,DC=zimmer,DC=com"));
+
+        var ex = Assert.Throws<LdapException>(() => NovellLdapConnection.ReadEntries(results));
+        Assert.Equal(LdapException.Busy, ex.ResultCode);
+    }
+
+    /// <summary>
+    /// A pathological client that throws a referral without advancing must not spin forever
+    /// inside a fail-closed authentication path.
+    /// </summary>
+    [Fact]
+    public void An_endless_referral_stream_terminates()
+    {
+        var results = new EndlessReferrals();
+
+        var entries = NovellLdapConnection.ReadEntries(results);
+
+        Assert.Empty(entries);
+        Assert.InRange(results.Calls, 1, 2_000);
+    }
+
+    // ---------------------------------------------------------------------
+
+    private sealed record Step(LdapEntry? Value, Exception? Throw)
+    {
+        public static Step Entry(string dn) => new(new LdapEntry(dn, new LdapAttributeSet()), null);
+        public static Step Referral() => new(null, new LdapReferralException("continuation reference"));
+        public static Step Throwing(Exception ex) => new(null, ex);
+    }
+
+    /// <summary>Replays a fixed script of entries and faults through the Novell result seam.</summary>
+    private sealed class ScriptedResults : ILdapSearchResults
+    {
+        private readonly Step[] _steps;
+        private int _index;
+
+        public ScriptedResults(params Step[] steps) => _steps = steps;
+
+        public int Count => _steps.Length;
+
+        public LdapControl[] ResponseControls => [];
+
+        public bool HasMore() => _index < _steps.Length;
+
+        public LdapEntry Next()
+        {
+            var step = _steps[_index++];
+            if (step.Throw is not null) throw step.Throw;
+            return step.Value!;
+        }
+
+        // The production reader drives HasMore()/Next() directly; enumeration is part of the
+        // interface but not part of the path under test.
+        public IEnumerator<LdapEntry> GetEnumerator() => throw new NotSupportedException();
+
+        System.Collections.IEnumerator System.Collections.IEnumerable.GetEnumerator() => GetEnumerator();
+    }
+
+    /// <summary>Always claims more results and always throws a referral, never advancing.</summary>
+    private sealed class EndlessReferrals : ILdapSearchResults
+    {
+        public int Calls { get; private set; }
+
+        public int Count => int.MaxValue;
+
+        public LdapControl[] ResponseControls => [];
+
+        public bool HasMore() => true;
+
+        public LdapEntry Next()
+        {
+            Calls++;
+            throw new LdapReferralException("continuation reference");
+        }
+
+        public IEnumerator<LdapEntry> GetEnumerator() => throw new NotSupportedException();
+
+        System.Collections.IEnumerator System.Collections.IEnumerable.GetEnumerator() => GetEnumerator();
+    }
+}
@@ -0,0 +1,51 @@
+using ZB.MOM.WW.Auth.Ldap.Internal;
+
+namespace ZB.MOM.WW.Auth.Ldap.Tests;
+
+/// <summary>
+/// Test double for <see cref="ILdapConnectionFactory"/> that hands out pre-scripted
+/// <see cref="FakeLdapConnection"/> instances in order, one per <see cref="Create"/> call —
+/// exactly matching the failover walk, which opens a fresh connection per endpoint attempt.
+/// Records every created connection for inspection. When the script runs out, returns a
+/// bare unscripted fake (which yields user-not-found) so a miscounted test fails loudly
+/// rather than reusing a stale fake.
+/// </summary>
+internal sealed class SequenceLdapConnectionFactory : ILdapConnectionFactory
+{
+    private readonly Queue<FakeLdapConnection> _script;
+
+    /// <summary>Scripts the connections handed out, in the order the walk will request them.</summary>
+    public SequenceLdapConnectionFactory(params FakeLdapConnection[] fakes)
+        => _script = new Queue<FakeLdapConnection>(fakes);
+
+    /// <summary>Every connection handed out, in creation order (scripted or overflow).</summary>
+    public List<FakeLdapConnection> Created { get; } = new();
+
+    /// <inheritdoc />
+    public ILdapConnection Create()
+    {
+        var fake = _script.Count > 0 ? _script.Dequeue() : new FakeLdapConnection();
+        Created.Add(fake);
+        return fake;
+    }
+}
+
+// ---------------------------------------------------------------------------
+// Smoke test: verifies the sequencing and overflow behaviour the failover tests rely on.
+// ---------------------------------------------------------------------------
+
+public class SequenceLdapConnectionFactorySmokeTests
+{
+    [Fact]
+    public void Create_ReturnsScriptedFakesInOrder_ThenOverflows()
+    {
+        var first = new FakeLdapConnection().ThrowOnConnect();
+        var second = new FakeLdapConnection();
+        var factory = new SequenceLdapConnectionFactory(first, second);
+
+        Assert.Same(first, factory.Create());
+        Assert.Same(second, factory.Create());
+        Assert.NotSame(second, factory.Create()); // overflow: fresh bare fake
+        Assert.Equal(3, factory.Created.Count);
+    }
+}
@@ -4,7 +4,7 @@ Startup configuration-validation library for the **ZB.MOM.WW SCADA family** (OtO

 The library normalizes the three-project configuration-validation surface: a failure-accumulating `IValidateOptions` base, reusable rule primitives, a bind+validate+`ValidateOnStart` DI extension, and a pre-host `ConfigPreflight` aggregator for raw `IConfiguration` — so the plumbing is written once and domain rules stay per-project.

-**Built at 0.1.0. Adopted by OtOpcUa, MxAccessGateway, and ScadaBridge on 2026-06-01** (local default branches; not yet pushed to remotes). Adoption tracked in `~/Desktop/scadaproj/components/configuration/GAPS.md`.
+**Built at 0.1.0. Adopted by OtOpcUa, MxAccessGateway, and ScadaBridge on 2026-06-01** (local default branches; not yet pushed to remotes — stale → #17). Adoption tracked in `~/Desktop/scadaproj/components/configuration/GAPS.md`.

 ---

@@ -66,7 +66,7 @@ ZB.MOM.WW.Configuration/

 ## Status

-Part of the **scadaproj component-normalization family** — this is the configuration + validation component. Built at **0.1.0**. **Adopted by OtOpcUa, MxAccessGateway, and ScadaBridge on 2026-06-01** (local default branches; not yet pushed to remotes) — per-app result is tracked in:
+Part of the **scadaproj component-normalization family** — this is the configuration + validation component. Built at **0.1.0**. **Adopted by OtOpcUa, MxAccessGateway, and ScadaBridge on 2026-06-01** (local default branches; not yet pushed to remotes — stale → #17) — per-app result is tracked in:

 - `~/Desktop/scadaproj/components/configuration/GAPS.md`

@@ -75,3 +75,33 @@ Design documentation:
 - `~/Desktop/scadaproj/components/configuration/spec/SPEC.md` — normalized validation target
 - `~/Desktop/scadaproj/components/configuration/shared-contract/ZB.MOM.WW.Configuration.md` — proposed shared-library API
 - `~/Desktop/scadaproj/components/configuration/current-state/` — per-project current state (code-verified)
+
+## Index history (migrated from the scadaproj root CLAUDE.md)
+
+> Moved verbatim out of the scadaproj root `CLAUDE.md` on 2026-09-03 so it loads only when needed. **This file is now the authoritative place for this history — append new entries here and keep the root index row short.**
+
+### Narrative
+
+The configuration component is fully populated: a normalized [`spec`](components/configuration/spec/SPEC.md), a
+[`shared-contract`](components/configuration/shared-contract/ZB.MOM.WW.Configuration.md), three
+[`current-state`](components/configuration/current-state/) docs, and an adoption [`GAPS`](components/configuration/GAPS.md)
+backlog. Shared = the `IValidateOptions<T>` failure-accumulation base (`OptionsValidatorBase<T>`) +
+reusable rule primitives (`ValidationBuilder`: port / host:port / required / positive-duration / one-of /
+min-count) + `AddValidatedOptions<TOptions,TValidator>()` (bind + validate + `ValidateOnStart`) + the
+pre-host `ConfigPreflight` aggregator (generalizes ScadaBridge's `StartupValidator`, byte-compatible
+message); left per-project = each app's options classes + domain rules, and OtOpcUa's
+draft/generation-content validation (DB-side `sp_ValidateDraft`; its C# `DraftValidator` is dormant).
+
+The shared library is **built and lives in this repo** at [`ZB.MOM.WW.Configuration/`](ZB.MOM.WW.Configuration/)
+(.NET 10; single package `ZB.MOM.WW.Configuration`; 42 tests (verified 2026-07-18); `dotnet pack` → 1 nupkg @ 0.1.0).
+The implementation plan is at
+[`docs/plans/2026-06-01-zb-mom-ww-configuration-shared-library.md`](docs/plans/2026-06-01-zb-mom-ww-configuration-shared-library.md).
+**Adopted across all three apps on 2026-06-01** (OtOpcUa, MxAccessGateway, ScadaBridge) on each repo's
+local default branch (`main`/`master`) — merged, **not yet pushed** to remotes (stale → #17); the package was first
+published to the Gitea feed. Behaviour-preserving onto `OptionsValidatorBase`/`AddValidatedOptions`
+for MxGateway + ScadaBridge (validator messages byte-identical), `StartupValidator` → `ConfigPreflight`
+for ScadaBridge, and net-new `Ldap`/`OpcUa` validators for OtOpcUa. Per-app result tracked in
+[`components/configuration/GAPS.md`](components/configuration/GAPS.md).
+Build/test from `ZB.MOM.WW.Configuration/`: `dotnet test`. Consumer matrix: all three apps consume the
+single package; ScadaBridge is the heaviest adopter (per-module validators + `StartupValidator` →
+`ConfigPreflight`); OtOpcUa adoption is additive (it has no `IValidateOptions` usage today).
@@ -0,0 +1,28 @@
+# ZB.MOM.WW.GalaxyRepository
+
+Shared Galaxy object-hierarchy SQL browse + `galaxy_repository.v1` gRPC service for the **ZB.MOM.WW SCADA family** (consumed by HistorianGateway and mxaccessgw). Build/test from this directory: `dotnet test`.
+
+## Index history (migrated from the scadaproj root CLAUDE.md)
+
+> Moved verbatim out of the scadaproj root `CLAUDE.md` on 2026-09-03 so it loads only when needed. **This file is now the authoritative place for this history — append new entries here and keep the root index row short.**
+
+### Narrative
+
+The Galaxy Repository component normalizes the **Galaxy object-hierarchy SQL browse + reusable gRPC service**
+that was previously embedded in `mxaccessgw`. Shared = canonical `galaxy_repository.v1` proto (wire-compatible
+with `mxaccessgw`'s existing contract so OtOpcUa's `GalaxyRepositoryClient` is unaffected), the SQL browse
+provider (`HierarchySql` / `AttributesSql` validated reverse-engineered queries), in-memory hierarchy cache +
+snapshot + deploy-poll refresh `BackgroundService`, `GalaxyHierarchyProjector`, and `AddZbGalaxyRepository` /
+`MapZbGalaxyRepository` DI extension. Left per-consumer = section path, subtree auth filtering, and any
+app-specific paging defaults.
+
+The shared library is **built and lives in this repo** at [`ZB.MOM.WW.GalaxyRepository/`](ZB.MOM.WW.GalaxyRepository/)
+(.NET 10; single package `ZB.MOM.WW.GalaxyRepository`; `dotnet pack` → 1 nupkg @ 0.1.0, **published to
+the Gitea NuGet feed** `gitea.dohertylan.com/api/packages/dohertj2/nuget`). The design doc is at
+[`docs/plans/2026-06-23-historian-gateway-design.md`](docs/plans/2026-06-23-historian-gateway-design.md) (§10, component 1).
+**Consumed by HistorianGateway** as a `PackageReference` from that Gitea feed, pinned at `0.2.0` (originally a
+cross-repo `ProjectReference` to this scadaproj tree; switched to the feed package 2026-06-24; bumped `0.1.0` → `0.2.0` 2026-06-25).
+**mxaccessgw adoption is a tracked follow-on** — once adopted, mxaccessgw's inline Galaxy browse code is replaced
+by the shared lib (the `galaxy_repository.v1` wire contract is already identical, so OtOpcUa and ScadaBridge
+clients are unaffected). Build/test from `ZB.MOM.WW.GalaxyRepository/`: `dotnet test`.
+Consumer matrix: HistorianGateway (initial); mxaccessgw (follow-on adoption).
@@ -2,9 +2,11 @@

 Health-check libraries for the **ZB.MOM.WW SCADA family** (OtOpcUa, MxAccessGateway, ScadaBridge). These are **libraries, not a service** — each package is linked directly into the consuming application at build time. There is no central health process or network hop; probes run in-process alongside the application.

+**Backlog / open work:** lives in Gitea issues on `dohertj2/scadaproj` (`tea issues list --repo dohertj2/scadaproj`) — currently #11 (re-verify per-app probe coverage vs the spec). The history below is a record, not a status tracker; file new follow-ups as issues.
+
 The library normalizes the three-tier health endpoint convention (`/health/ready`, `/health/active`, `/healthz`) and provides reusable probe implementations so the three sister projects share a common surface without duplicating probe logic.

-**Built at 0.1.0. NOT yet adopted by the three apps.** Adoption is tracked in `~/Desktop/scadaproj/components/health/GAPS.md`.
+**Built at 0.3.0 and published to the Gitea NuGet feed. Adopted by all four apps** — adoption is tracked in `~/Desktop/scadaproj/components/health/GAPS.md`.

 ---

@@ -13,7 +15,7 @@ The library normalizes the three-tier health endpoint convention (`/health/ready
 | Package | Responsibilities | Key Dependencies |
 |---|---|---|
 | `ZB.MOM.WW.Health` | Core tier convention, `MapZbHealth` extension, canonical JSON writer (`ZbHealthWriter`), `IActiveNodeGate` seam, `GrpcDependencyHealthCheck` reachability probe, tier-tag constants (`ZbHealthTags`). No Akka or EF dependency. | `Microsoft.AspNetCore.App` (framework ref), `Grpc.Net.Client` |
-| `ZB.MOM.WW.Health.Akka` | `AkkaClusterHealthCheck` with a configurable `AkkaClusterStatusPolicy` (presets: `Default` / `OtOpcUaCompat`), `ActiveNodeHealthCheck` with an optional role filter, and `AkkaActiveNodeGate` that backs `IActiveNodeGate` from cluster member state. | `ZB.MOM.WW.Health`, `Akka.Cluster` |
+| `ZB.MOM.WW.Health.Akka` | `AkkaClusterHealthCheck` with a configurable `AkkaClusterStatusPolicy` (presets: `Default` / `OtOpcUaCompat`); `ClusterActiveNode` — the family-wide **oldest-`Up`-member** rule, optionally role-scoped; `ActiveNodeHealthCheck` over it (role-preference list, `activeRole`/`activeNode` in the result `data`); and `AkkaActiveNodeGate`, which backs `IActiveNodeGate` from the same rule. | `ZB.MOM.WW.Health`, `Akka.Cluster` |
 | `ZB.MOM.WW.Health.EntityFrameworkCore` | `DatabaseHealthCheck<TContext>` — resolves `IDbContextFactory<TContext>` when registered, else a scoped `TContext`; pool-safe. Default probe: `CanConnectAsync`. Optional `ProbeQuery` delegate for query-based validation. | `ZB.MOM.WW.Health`, `Microsoft.EntityFrameworkCore` |

 ---
@@ -46,14 +48,19 @@ dotnet test ZB.MOM.WW.Health.slnx
 dotnet pack ZB.MOM.WW.Health.slnx -c Release -o ./artifacts
 ```

-All three test assemblies run offline:
+All three test assemblies run offline. The Akka suite forms real in-process clusters on loopback: a
+single-node one for the cluster-data payload, and a **two-node one built so the oldest member is not
+the lowest-addressed one** — the only condition under which age- and leader-ordering disagree, and so
+the only fixture that can catch a leader-based active-node regression. It binds fixed ports
+19540/19541 (distinct from OtOpcUa's equivalent fixture) and carries a sanity check that the
+divergence actually occurred, so the assertions cannot pass for the wrong reason:

 | Assembly | Tests |
 |---|---|
-| `ZB.MOM.WW.Health.Tests` | 20 |
-| `ZB.MOM.WW.Health.Akka.Tests` | 32 |
+| `ZB.MOM.WW.Health.Tests` | 25 |
+| `ZB.MOM.WW.Health.Akka.Tests` | 45 |
 | `ZB.MOM.WW.Health.EntityFrameworkCore.Tests` | 6 |
-| **Total** | **58** |
+| **Total** | **76** |

 `GeneratePackageOnBuild` is off — pack explicitly with the command above.

@@ -61,7 +68,7 @@ All three test assemblies run offline:

 ## Status

-Built at **0.1.0** and published to the Gitea NuGet feed. **Not yet adopted** by the three apps — adoption is tracked in the component backlog:
+Built at **0.3.0** and published to the Gitea NuGet feed. **Adopted** by all four apps — adoption is tracked in the component backlog:

 - `~/Desktop/scadaproj/components/health/GAPS.md` — adoption order, effort, and risk

@@ -70,3 +77,30 @@ Design documentation:
 - `~/Desktop/scadaproj/components/health/spec/SPEC.md` — normalized three-tier target
 - `~/Desktop/scadaproj/components/health/shared-contract/ZB.MOM.WW.Health.md` — proposed API (aligned to shipped code)
 - `~/Desktop/scadaproj/components/health/current-state/` — per-project current state (code-verified)
+
+## Index history (migrated from the scadaproj root CLAUDE.md)
+
+> Moved verbatim out of the scadaproj root `CLAUDE.md` on 2026-09-03 so it loads only when needed. **This file is now the authoritative place for this history — append new entries here and keep the root index row short.**
+
+### Component-table row as it read in the index
+
+| Component | Status | Goal | Design | Implementation |
+|---|---|---|---|---|
+| Health (readiness / liveness / active-node) | **Adopted** (feed carries `0.1.0`–`0.3.0`, current lib version **`0.3.0`**; **all 4 apps** reference it + wire `MapZbHealth`). **`0.2.0` (2026-07-24)** added an optional per-entry `data` object to the canonical health JSON and made `AkkaClusterHealthCheck` publish its cluster view (`leader`/`selfAddress`/`selfRoles`/`memberCount`/`unreachableCount`) — additive, so pre-0.2.0 payloads still parse. **`0.2.1` (2026-07-24) is a behaviour fix with a live consumer impact:** the active tier now returns **Unhealthy (503)** on a role-member-but-not-leader node instead of `Degraded`, which the spec maps to 200. The old value made `/health/active` answer 200 on *every* node, so it could not distinguish active from standby over HTTP at all — and OtOpcUa's Traefik routes the admin UI by exactly this probe, so its leader-pinning had silently never worked ([`lmxopcua#494`](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/494)). **`0.3.0` (2026-07-24) finished that fix and ENDED a family-wide duplication.** 0.2.1 made the tier a real 503 but still selected by `RoleLeader`; on a rebuilt rig the two orderings diverge live (akka leader `central-1`, oldest admin `central-2` — the singleton host), so it would have pinned Traefik to the node *not* hosting the work. **Leadership is address-ordered (host, then port) and has no relationship to time; singleton placement is age-ordered.** They agree on a freshly-formed cluster — which is why single-node and happy-path tests never caught it — and diverge permanently after any restart. Both Akka apps had already discovered this independently and each written a private replacement rather than use the shared check (ScadaBridge `OldestNodeActiveHealthCheck`, OtOpcUa `ClusterPrimaryHealthCheck`), so the package's entire active-node surface had **zero consumers — not unused but avoided, twice, for the same reason.** 0.3.0 promotes those two into one `ClusterActiveNode` primitive (oldest `Up` member, optional role scope, plus an ordered role-preference so a fused node answers for the role its singletons are pinned to); `ActiveNodeHealthCheck` and `AkkaActiveNodeGate` both delegate to it, and results now carry `activeRole`/`selfAddress`/`activeNode` in the 0.2.0 `data` object so a **standby reports who IS active**. Two further behaviour changes: a node holding **none** of the preferred roles is now **Unhealthy**, not Healthy — the old "not applicable ⇒ Healthy" is exactly what made the tier answer 200 everywhere, and the case is reachable (OtOpcUa's `RoleParser` admits `dev`-only and `cluster-`only nodes); and identity compares `UniqueAddress`, so a node restarted on the same host:port is correctly a different member. **Both apps migrated and their private copies deleted** — OtOpcUa (`feat/health-0.3.0-active-node` @ `88adec04`) also routes `RedundancyStateActor.SelectOldestUpMemberOfRole` through the shared primitive, so the tier and the OPC UA `ServiceLevel` 250/240 split cannot drift; ScadaBridge (`feat/site-node-health` @ `5d4853a1`) routes `ActiveNodeEvaluator` through it, unifying the S&F delivery gate, the heartbeat `IsActive` stamp, the inbound-API gate and the health tier on one rule. ScadaBridge keeps `SitePairActiveNodeHealthCheck` deliberately — it is a thin adapter over the site's own role-scoped `IClusterNodeProvider` (itself now backed by the shared rule), not a second copy of it. mxgw/HistorianGateway have no Akka cluster and are unaffected. **Both app branches are MERGED and PUSHED** (verified 2026-08-07): OtOpcUa `88adec04` is on `origin/master`; ScadaBridge merged via `b3dc17a5` (2026-07-24) on `origin/main`, private-copy deletions included. | Shared `ZB.MOM.WW.Health` lib | [`components/health/`](components/health/) | [`ZB.MOM.WW.Health/`](ZB.MOM.WW.Health/) |
+
+### Narrative
+
+The health component is fully populated: a normalized [`spec`](components/health/spec/SPEC.md), a
+[`shared-contract`](components/health/shared-contract/ZB.MOM.WW.Health.md), three
+[`current-state`](components/health/current-state/) docs, and an adoption [`GAPS`](components/health/GAPS.md)
+backlog. Shared = three-tier endpoint convention (ready/active/healthz) + canonical JSON writer +
+`IActiveNodeGate` seam + `GrpcDependencyHealthCheck` + `AkkaClusterHealthCheck` + `ActiveNodeHealthCheck`
+ `DatabaseHealthCheck<TContext>`; left per-project = which probes each app registers,
+orchestrator wiring, and ScadaBridge's distributed health-monitoring pipeline.
+
+The shared library is **built and lives in this repo** at [`ZB.MOM.WW.Health/`](ZB.MOM.WW.Health/)
+(.NET 10; 3 packages — `ZB.MOM.WW.Health`, `ZB.MOM.WW.Health.Akka`, `ZB.MOM.WW.Health.EntityFrameworkCore`;
+64 tests (verified 2026-07-18); `dotnet pack` → 3 nupkgs @ 0.1.0).
+**Adopted by all four apps** — verified 2026-07-18 (each references the package and wires `MapZbHealth`); the previous "not yet adopted" claim was stale. Per-app probe coverage vs the spec has NOT been re-verified (→ #11) — see [`components/health/GAPS.md`](components/health/GAPS.md).
+Build/test from `ZB.MOM.WW.Health/`: `dotnet test`. Consumer matrix: MxAccessGateway → core only;
+OtOpcUa & ScadaBridge → all three packages.
@@ -5,7 +5,7 @@
    <Nullable>enable</Nullable>
    <ImplicitUsings>enable</ImplicitUsings>
    <LangVersion>latest</LangVersion>
-    <Version>0.1.0</Version>
+    <Version>0.3.0</Version>
    <ManagePackageVersionsCentrally>true</ManagePackageVersionsCentrally>
    <!-- Emit XML docs so the public API summaries ship inside the packed nupkgs (IntelliSense for
         consumers). CS1591 (missing doc on a public member) is suppressed so undocumented test /
@@ -30,7 +30,19 @@ MxAccessGateway consumes the core package only — it has no Akka cluster and no

 ## Versioning

-All three packages are versioned **lockstep** from `Directory.Build.props`. The current release is **0.1.0**. A single version bump in `Directory.Build.props` bumps all three packages simultaneously — consumers should reference the same version for all ZB.MOM.WW.Health packages.
+All three packages are versioned **lockstep** from `Directory.Build.props`. The current release is **0.2.0**. A single version bump in `Directory.Build.props` bumps all three packages simultaneously — consumers should reference the same version for all ZB.MOM.WW.Health packages.
+
+### 0.2.0 — per-entry `data` (additive, non-breaking)
+
+- `ZbHealthWriter` emits an optional `"data": { … }` object per entry, sourced from the check's
+  `HealthCheckResult.Data`, **only when non-empty** — a check that publishes no data produces a
+  byte-identical body to 0.1.0, so every existing consumer is unaffected. Data keys are written
+  verbatim (the camelCase policy applies to the envelope, not to dictionary keys).
+- `AkkaClusterHealthCheck` populates that dictionary with this node's cluster view: `leader`
+  (omitted while unknown), `selfAddress`, `selfRoles`, `memberCount`, `unreachableCount`. The
+  startup-safety paths (no `ActorSystem`, cluster not yet accessible) stay description-only.
+- Consumers gain the leader field on a package bump alone — no application code change, provided the
+  app registers the shared `AkkaClusterHealthCheck`.

 ---

@@ -42,14 +54,16 @@ dotnet build ZB.MOM.WW.Health.slnx
 dotnet test  ZB.MOM.WW.Health.slnx
 ```

-All three test assemblies run with `dotnet test` and require no external dependencies (no running Akka cluster, no database):
+All three test assemblies run with `dotnet test` and require no external dependencies — no database and no
+external cluster (the cluster-data tests form a single-node in-process Akka cluster on a loopback port
+via `Akka.TestKit.Xunit2`):

 | Assembly | Tests |
 |---|---|
-| `ZB.MOM.WW.Health.Tests` | 20 |
-| `ZB.MOM.WW.Health.Akka.Tests` | 32 |
+| `ZB.MOM.WW.Health.Tests` | 25 |
+| `ZB.MOM.WW.Health.Akka.Tests` | 39 |
 | `ZB.MOM.WW.Health.EntityFrameworkCore.Tests` | 6 |
-| **Total** | **58** |
+| **Total** | **70** |

 ---

@@ -62,9 +76,9 @@ dotnet pack ZB.MOM.WW.Health.slnx -c Release -o ./artifacts
 Produces three `.nupkg` files in `artifacts/`:

 ```
-ZB.MOM.WW.Health.0.1.0.nupkg
-ZB.MOM.WW.Health.Akka.0.1.0.nupkg
-ZB.MOM.WW.Health.EntityFrameworkCore.0.1.0.nupkg
+ZB.MOM.WW.Health.0.2.0.nupkg
+ZB.MOM.WW.Health.Akka.0.2.0.nupkg
+ZB.MOM.WW.Health.EntityFrameworkCore.0.2.0.nupkg
 ```

 `GeneratePackageOnBuild` is off — pack explicitly as above.
@@ -73,7 +87,7 @@ ZB.MOM.WW.Health.EntityFrameworkCore.0.1.0.nupkg

 ## Status

-**Built at 0.1.0. NOT yet adopted by the three apps.** Adoption is tracked in the component backlog:
+**Built at 0.2.0 and published to the Gitea NuGet feed. Adopted by all four apps** (OtOpcUa, MxAccessGateway, ScadaBridge, HistorianGateway). Adoption is tracked in the component backlog:

 - `~/Desktop/scadaproj/components/health/GAPS.md`

@@ -9,89 +9,80 @@ using Microsoft.Extensions.Diagnostics.HealthChecks;
 namespace ZB.MOM.WW.Health.Akka;

 /// <summary>
-/// Pure decision function for the active / leader probe, factored out of
-/// <see cref="ActiveNodeHealthCheck"/> so the role-less and role-filtered matrices are exhaustively
-/// table-testable without forming a real cluster.
-/// </summary>
-internal static class ActiveNodeDecision
-{
-    /// <summary>
-    /// Maps the resolved cluster facts to a <see cref="HealthStatus"/>.
-    /// </summary>
-    /// <param name="selfUp">Whether the local node's member status is <c>Up</c>.</param>
-    /// <param name="isLeader">
-    /// Whether the local node is the leader: the cluster leader in role-less mode, or the
-    /// role-singleton leader in role-filtered mode.
-    /// </param>
-    /// <param name="hasRole">
-    /// Whether the local node carries <paramref name="requiredRole"/>. Ignored when
-    /// <paramref name="requiredRole"/> is <c>null</c>.
-    /// </param>
-    /// <param name="requiredRole">
-    /// The role to scope the check to, or <c>null</c> for the role-less (whole-cluster-leader) mode.
-    /// </param>
-    /// <returns>
-    /// Role-less: Healthy iff the node is Up and the cluster leader, otherwise Unhealthy.
-    /// Role-filtered: Healthy when the node lacks the role (probe irrelevant) or carries the role and
-    /// is the role-singleton leader; Degraded when it carries the role but is not the leader.
-    /// </returns>
-    public static HealthStatus Evaluate(bool selfUp, bool isLeader, bool hasRole, string? requiredRole)
-    {
-        if (requiredRole is null)
-            return selfUp && isLeader ? HealthStatus.Healthy : HealthStatus.Unhealthy;
-
-        if (!hasRole)
-            return HealthStatus.Healthy;
-
-        return isLeader ? HealthStatus.Healthy : HealthStatus.Degraded;
-    }
-}
-
-/// <summary>
-/// Health check that reports whether this node is the designated active / leader node.
-/// An optional role scopes the check to nodes carrying that role. Register to the
-/// <see cref="ZbHealthTags.Active"/> tag.
+/// Active-tier probe: Healthy (200) on the one node in charge, Unhealthy (503) on its standby.
+/// Register to the <see cref="ZbHealthTags.Active"/> tag.
 /// </summary>
 /// <remarks>
-/// The <see cref="ActorSystem"/> is resolved lazily from the service provider. If it is not yet
-/// available — e.g. during startup before Akka is initialised — the check returns
-/// <see cref="HealthStatus.Degraded"/> rather than throwing, so it is startup-safe.
+///     <para>
+///         "In charge" means the <b>oldest Up member</b> — see <see cref="ClusterActiveNode"/> for
+///         why leadership is the wrong answer and what breaks when it is used. Scope the competition
+///         to a role with <see cref="ActiveNodeHealthCheckOptions.RolePreference"/>; leave it empty
+///         for a cluster whose nodes all compete for the same single active slot.
+///     </para>
+///     <para>
+///         <b>Behaviour change in 0.3.0.</b> Through 0.2.1 this check selected by
+///         <c>ClusterState.Leader</c> / <c>RoleLeader</c>, and its role-filtered mode reported
+///         Healthy for any node <i>lacking</i> the role. Both were wrong, both were found on live
+///         rigs, and both Akka consumers had already replaced it with a private copy rather than use
+///         it. It now selects by age and reports <see cref="HealthStatus.Unhealthy"/> for a node that
+///         owns no active work. A node whose active tier previously answered 200 unconditionally will
+///         now answer 503 unless it really is the active node — which is the point.
+///     </para>
+///     <para>
+///         Startup-safe: before the <see cref="ActorSystem"/> or the cluster is available the result
+///         is <see cref="HealthStatus.Degraded"/> (a 200 under the tier mapping), so a node that is
+///         merely still booting is not reported as a failed standby and pulled from rotation on the
+///         way up.
+///     </para>
 /// </remarks>
 public sealed class ActiveNodeHealthCheck : IHealthCheck
 {
    private readonly IServiceProvider _serviceProvider;
-    private readonly string? _role;
+    private readonly ActiveNodeHealthCheckOptions _options;

    /// <summary>
-    /// Role-less constructor: Healthy when the node is <c>Up</c> and the cluster leader
-    /// (ScadaBridge ActiveNode pattern); Unhealthy otherwise. Degraded when the ActorSystem /
-    /// cluster is not yet ready.
+    /// Unscoped: the active node is the oldest Up member of the whole cluster.
    /// </summary>
    /// <param name="serviceProvider">
    /// The application service provider. The <see cref="ActorSystem"/> is resolved lazily so the
-    /// check is startup-safe: if no <see cref="ActorSystem"/> is registered yet the result is Degraded.
+    /// check is startup-safe.
    /// </param>
    public ActiveNodeHealthCheck(IServiceProvider serviceProvider)
+        : this(serviceProvider, new ActiveNodeHealthCheckOptions())
    {
-        _serviceProvider = serviceProvider ?? throw new ArgumentNullException(nameof(serviceProvider));
-        _role = null;
    }

    /// <summary>
-    /// Role-filtered constructor: Healthy when the node lacks <paramref name="role"/> or carries it
-    /// and is the role-singleton leader; Degraded when it carries the role but is not the leader
-    /// (OtOpcUa AdminRoleLeader pattern). Degraded when the ActorSystem / cluster is not yet ready.
+    /// Role-scoped: the active node is the oldest Up member carrying <paramref name="role"/>.
    /// </summary>
    /// <param name="serviceProvider">
    /// The application service provider. The <see cref="ActorSystem"/> is resolved lazily so the
-    /// check is startup-safe: if no <see cref="ActorSystem"/> is registered yet the result is Degraded.
+    /// check is startup-safe.
    /// </param>
-    /// <param name="role">The Akka cluster role to scope the check to.</param>
+    /// <param name="role">The cluster role to scope the competition to.</param>
    public ActiveNodeHealthCheck(IServiceProvider serviceProvider, string role)
+        : this(serviceProvider, NewOptions(role))
+    {
+    }
+
+    /// <summary>
+    /// Configured with an explicit <see cref="ActiveNodeHealthCheckOptions"/>.
+    /// </summary>
+    /// <param name="serviceProvider">
+    /// The application service provider. The <see cref="ActorSystem"/> is resolved lazily so the
+    /// check is startup-safe.
+    /// </param>
+    /// <param name="options">The role-scoping options.</param>
+    public ActiveNodeHealthCheck(IServiceProvider serviceProvider, ActiveNodeHealthCheckOptions options)
    {
        _serviceProvider = serviceProvider ?? throw new ArgumentNullException(nameof(serviceProvider));
+        _options = options ?? throw new ArgumentNullException(nameof(options));
+    }
+
+    private static ActiveNodeHealthCheckOptions NewOptions(string role)
+    {
        ArgumentException.ThrowIfNullOrWhiteSpace(role);
-        _role = role;
+        return new ActiveNodeHealthCheckOptions { RolePreference = new[] { role } };
    }

    /// <inheritdoc />
@@ -103,10 +94,8 @@ public sealed class ActiveNodeHealthCheck : IHealthCheck
        if (system is null)
            return Task.FromResult(HealthCheckResult.Degraded("ActorSystem not yet available."));

-        MemberStatus selfStatus;
-        bool selfUp;
-        bool hasRole;
-        bool isLeader;
+        Member self;
+        IEnumerable<Member> members;
        try
        {
            // Reading cluster membership can throw while the ActorSystem exists but the cluster has
@@ -114,54 +103,49 @@ public sealed class ActiveNodeHealthCheck : IHealthCheck
            // ConfigurationException). The spec's startup-safety rule maps this to Degraded rather
            // than letting the exception escape (which the host would record as Unhealthy).
            var cluster = Cluster.Get(system);
-            var self = cluster.SelfMember;
-            selfStatus = self.Status;
-            selfUp = selfStatus == MemberStatus.Up;
-
-            if (_role is null)
-            {
-                hasRole = false;
-                var leader = cluster.State.Leader;
-                isLeader = leader is not null && leader == self.Address;
-            }
-            else
-            {
-                hasRole = self.HasRole(_role);
-                var roleLeader = cluster.State.RoleLeader(_role);
-                isLeader = roleLeader is not null && roleLeader == self.Address;
-            }
+            self = cluster.SelfMember;
+            members = cluster.State.Members;
        }
        catch (Exception ex) when (ex is not OperationCanceledException)
        {
            return Task.FromResult(HealthCheckResult.Degraded("Akka cluster state not yet accessible.", ex));
        }

-        var health = ActiveNodeDecision.Evaluate(selfUp, isLeader, hasRole, _role);
-        var description = DescribeResult(health, selfStatus, selfUp, isLeader);
-        var result = health switch
-        {
-            HealthStatus.Healthy   => HealthCheckResult.Healthy(description),
-            HealthStatus.Degraded  => HealthCheckResult.Degraded(description),
-            _                      => HealthCheckResult.Unhealthy(description),
-        };
-        return Task.FromResult(result);
-    }
+        var scoped = _options.RolePreference.Count > 0;
+        var role = ClusterActiveNode.ResolveActiveRole(self.Roles, _options.RolePreference);

-    private string DescribeResult(HealthStatus health, MemberStatus status, bool selfUp, bool isLeader)
-    {
-        if (_role is null)
+        var data = new Dictionary<string, object>(StringComparer.Ordinal)
        {
-            if (health == HealthStatus.Healthy)
-                return "Active node (cluster leader).";
-            return selfUp && !isLeader
-                ? "Standby: node is Up but not the cluster leader."
-                : $"Standby: node is not Up (status: {status}).";
+            ["selfAddress"] = self.Address.ToString(),
+        };
+        if (role is not null) data["activeRole"] = role;
+
+        if (scoped && role is null)
+        {
+            // The node carries none of the preferred roles, so it owns no active work and cannot be
+            // the active node for any of them. Reported per NoActiveRoleStatus — Unhealthy by
+            // default, because the active tier's whole job is to keep such a node out of the pool.
+            var message = $"Node carries none of the active roles: {string.Join(", ", _options.RolePreference)}.";
+            return Task.FromResult(new HealthCheckResult(_options.NoActiveRoleStatus, message, exception: null, data));
        }

-        return health switch
+        var active = ClusterActiveNode.OldestUpMember(members, role);
+        if (active is not null) data["activeNode"] = active.Address.ToString();
+
+        var scopeLabel = role is null ? "the cluster" : $"role '{role}'";
+
+        if (active is null)
        {
-            HealthStatus.Healthy => $"Active for role '{_role}' (or not a role member).",
-            _ => $"Role '{_role}' member but not leader.",
-        };
+            // Nobody is Up in the scope — a cold start, or every holder is still Joining or already
+            // Leaving. Nobody is active, so nobody may claim to be.
+            return Task.FromResult(HealthCheckResult.Unhealthy(
+                $"No Up member currently holds the active slot for {scopeLabel}.", exception: null, data));
+        }
+
+        // Compared by UniqueAddress: a node that restarted on the same host and port is a different
+        // member, and during the overlap the cluster briefly holds both incarnations.
+        return Task.FromResult(active.UniqueAddress.Equals(self.UniqueAddress)
+            ? HealthCheckResult.Healthy($"Active node for {scopeLabel} (oldest Up member).", data)
+            : HealthCheckResult.Unhealthy($"Standby: active node for {scopeLabel} is {active.Address}.", exception: null, data));
    }
 }
@@ -0,0 +1,37 @@
+using Microsoft.Extensions.Diagnostics.HealthChecks;
+
+namespace ZB.MOM.WW.Health.Akka;
+
+/// <summary>
+/// Configures how <see cref="ActiveNodeHealthCheck"/> decides which role's active node this node is
+/// competing to be.
+/// </summary>
+public sealed class ActiveNodeHealthCheckOptions
+{
+    /// <summary>
+    /// Candidate roles in descending precedence. The check answers for the first one the local node
+    /// carries. Empty (the default) means unscoped: the active node is the oldest Up member of the
+    /// whole cluster.
+    /// </summary>
+    /// <remarks>
+    /// A fused node should list the role its singletons are pinned to first — e.g.
+    /// <c>["admin", "driver"]</c> for a node that hosts the admin singletons when fused and competes
+    /// as a plain driver when it is a driver-only node.
+    /// </remarks>
+    public IReadOnlyList<string> RolePreference { get; init; } = Array.Empty<string>();
+
+    /// <summary>
+    /// The status to report when <see cref="RolePreference"/> is non-empty and the node carries none
+    /// of those roles — it owns no active work at all. Defaults to
+    /// <see cref="HealthStatus.Unhealthy"/>.
+    /// </summary>
+    /// <remarks>
+    /// Unhealthy is the safe default and is deliberately <b>not</b> the "not applicable ⇒ Healthy"
+    /// convention used by readiness probes. The active tier exists so a load balancer can pick
+    /// exactly one node; a node that owns no active work must not be in that pool. Answering Healthy
+    /// there is what made the previous role-filtered check report 200 on every node and silently
+    /// broke leader-pinning (lmxopcua#494). Override only when a node genuinely should stay in the
+    /// active pool without holding any of the preferred roles.
+    /// </remarks>
+    public HealthStatus NoActiveRoleStatus { get; init; } = HealthStatus.Unhealthy;
+}
@@ -6,27 +6,47 @@ namespace ZB.MOM.WW.Health.Akka;

 /// <summary>
 /// <see cref="IActiveNodeGate"/> implementation that computes <see cref="IsActiveNode"/> directly
-/// from the Akka cluster state (self member <c>Up</c> and the local node is the cluster leader).
-/// Register as a singleton.
+/// from the Akka cluster state — the local node is the <b>oldest Up member</b>, optionally within a
+/// role scope. Register as a singleton.
 /// </summary>
 /// <remarks>
-/// The <see cref="ActorSystem"/> is resolved lazily from the service provider; if it is not yet
-/// available — e.g. during startup before Akka is initialised — <see cref="IsActiveNode"/> returns
-/// <c>false</c> (the safe default during startup). This gate reads the cluster state directly and
-/// does not resolve <see cref="ActiveNodeHealthCheck"/> from DI.
+///     <para>
+///         Shares its rule with <see cref="ActiveNodeHealthCheck"/> via
+///         <see cref="ClusterActiveNode"/>, so an endpoint gated by this type and the
+///         <c>/health/active</c> probe an orchestrator routes by can never disagree about which node
+///         is in charge.
+///     </para>
+///     <para>
+///         <b>Behaviour change in 0.3.0:</b> through 0.2.1 this gate selected by
+///         <c>ClusterState.Leader</c>, which is address-ordered and diverges from singleton placement
+///         after any restart. See <see cref="ClusterActiveNode"/>.
+///     </para>
+///     <para>
+///         The <see cref="ActorSystem"/> is resolved lazily from the service provider; if it is not
+///         yet available — e.g. during startup before Akka is initialised —
+///         <see cref="IsActiveNode"/> returns <c>false</c> (the safe default during startup, matching
+///         the standby case). This gate reads the cluster state directly and does not resolve
+///         <see cref="ActiveNodeHealthCheck"/> from DI.
+///     </para>
 /// </remarks>
 public sealed class AkkaActiveNodeGate : IActiveNodeGate
 {
    private readonly IServiceProvider _serviceProvider;
+    private readonly string? _role;

    /// <summary>Initializes a new <see cref="AkkaActiveNodeGate"/>.</summary>
    /// <param name="serviceProvider">
    /// The application service provider. The <see cref="ActorSystem"/> is resolved lazily; if it is
    /// not yet available <see cref="IsActiveNode"/> returns <c>false</c>.
    /// </param>
-    public AkkaActiveNodeGate(IServiceProvider serviceProvider)
+    /// <param name="role">
+    /// The role to scope the competition to, or <c>null</c> (the default) to consider every Up
+    /// member.
+    /// </param>
+    public AkkaActiveNodeGate(IServiceProvider serviceProvider, string? role = null)
    {
        _serviceProvider = serviceProvider ?? throw new ArgumentNullException(nameof(serviceProvider));
+        _role = role;
    }

    /// <inheritdoc />
@@ -38,13 +58,16 @@ public sealed class AkkaActiveNodeGate : IActiveNodeGate
            if (system is null)
                return false;

-            var cluster = Cluster.Get(system);
-            var self = cluster.SelfMember;
-            if (self.Status != MemberStatus.Up)
+            try
+            {
+                return ClusterActiveNode.SelfIsActive(Cluster.Get(system), _role);
+            }
+            catch (Exception)
+            {
+                // ActorSystem exists but the cluster is not initialised yet. Same startup-safety
+                // rule as the health check, expressed as the gate's safe default: not active.
                return false;
-
-            var leader = cluster.State.Leader;
-            return leader is not null && leader == self.Address;
+            }
        }
    }
 }
@@ -44,6 +44,7 @@ public sealed class AkkaClusterHealthCheck : IHealthCheck
            return Task.FromResult(HealthCheckResult.Degraded("ActorSystem not yet available."));

        MemberStatus status;
+        IReadOnlyDictionary<string, object> data;
        try
        {
            // Cluster.Get(system).SelfMember can throw while the ActorSystem exists but the cluster
@@ -51,7 +52,9 @@ public sealed class AkkaClusterHealthCheck : IHealthCheck
            // ConfigurationException). The spec's startup-safety rule maps this to Degraded, not an
            // escaping exception (which the host would record as Unhealthy and pull the node from
            // rotation).
-            status = Cluster.Get(system).SelfMember.Status;
+            var cluster = Cluster.Get(system);
+            status = cluster.SelfMember.Status;
+            data = BuildClusterData(cluster);
        }
        catch (Exception ex) when (ex is not OperationCanceledException)
        {
@@ -62,10 +65,47 @@ public sealed class AkkaClusterHealthCheck : IHealthCheck
        var description = $"Akka cluster member status: {status}";
        var result = health switch
        {
-            HealthStatus.Healthy   => HealthCheckResult.Healthy(description),
-            HealthStatus.Degraded  => HealthCheckResult.Degraded(description),
-            _                      => HealthCheckResult.Unhealthy(description),
+            HealthStatus.Healthy   => HealthCheckResult.Healthy(description, data),
+            HealthStatus.Degraded  => HealthCheckResult.Degraded(description, data: data),
+            _                      => HealthCheckResult.Unhealthy(description, data: data),
        };
        return Task.FromResult(result);
    }
+
+    /// <summary>
+    /// Projects this node's view of the cluster into the check's <c>data</c> dictionary.
+    /// </summary>
+    /// <param name="cluster">The cluster extension for the local <see cref="ActorSystem"/>.</param>
+    /// <returns>
+    /// JSON-friendly values only (string / int / string[]), so the canonical health writer can emit
+    /// them verbatim: <c>leader</c> (omitted while the leader is unknown, e.g. pre-join or during
+    /// convergence), <c>selfAddress</c>, <c>selfRoles</c> (sorted for a stable payload),
+    /// <c>memberCount</c>, <c>unreachableCount</c>.
+    /// </returns>
+    /// <remarks>
+    /// This is deliberately each node's <em>own</em> view rather than an authoritative fleet answer —
+    /// a consumer that reads it from several members of the same cluster can compare them, and a
+    /// disagreement is a split-brain tell.
+    /// </remarks>
+    internal static IReadOnlyDictionary<string, object> BuildClusterData(Cluster cluster)
+    {
+        ArgumentNullException.ThrowIfNull(cluster);
+
+        var state = cluster.State;
+        var data = new Dictionary<string, object>(StringComparer.Ordinal)
+        {
+            ["selfAddress"] = cluster.SelfAddress.ToString(),
+            ["selfRoles"] = cluster.SelfMember.Roles.OrderBy(static r => r, StringComparer.Ordinal).ToArray(),
+            ["memberCount"] = state.Members.Count,
+            ["unreachableCount"] = state.Unreachable.Count,
+        };
+
+        // Omitted rather than emitted as null: a missing key reads as "this node does not know yet",
+        // which is exactly the pre-join / mid-convergence state.
+        var leader = state.Leader;
+        if (leader is not null)
+            data["leader"] = leader.ToString();
+
+        return data;
+    }
 }
@@ -0,0 +1,133 @@
+using Akka.Cluster;
+
+namespace ZB.MOM.WW.Health.Akka;
+
+/// <summary>
+/// THE family-wide definition of "the active node": the <b>oldest Up member</b> of the cluster,
+/// optionally scoped to a role — the member <c>ClusterSingletonManager</c> places singletons on.
+/// </summary>
+/// <remarks>
+///     <para>
+///         Cluster <b>leadership</b> (<c>ClusterState.Leader</c> / <c>RoleLeader</c>) is the wrong
+///         answer and must never be used for a product-level active/standby decision. Leadership is
+///         <i>address</i>-ordered (host, then port) and has no relationship to time; singleton
+///         placement is <i>age</i>-ordered. On a freshly-formed cluster the two agree, which is why
+///         the discrepancy hides in unit tests and in a rig that has never been restarted. They
+///         diverge permanently after any restart: the restarted node rejoins as the youngest but
+///         keeps its address, so if it holds the lower address it becomes leader while the
+///         singletons — and all the work they own — stay on the other node.
+///     </para>
+///     <para>
+///         Both Akka consumers in the family independently hit this and each wrote its own
+///         replacement rather than use the shared leader-based check
+///         (ScadaBridge <c>OldestNodeActiveHealthCheck</c>, OtOpcUa <c>ClusterPrimaryHealthCheck</c>).
+///         This type is those two implementations promoted to one, so the health tier, the inbound
+///         API gate, the store-and-forward delivery gate and OtOpcUa's OPC UA <c>ServiceLevel</c>
+///         cannot drift apart about which node is in charge.
+///     </para>
+///     <para>
+///         Only <see cref="MemberStatus.Up"/> members are eligible, matching singleton placement: a
+///         <c>Leaving</c> node is handing its singletons over and must not be named active.
+///     </para>
+/// </remarks>
+public static class ClusterActiveNode
+{
+    /// <summary>
+    /// The oldest <see cref="MemberStatus.Up"/> member, optionally scoped to <paramref name="role"/>.
+    /// </summary>
+    /// <param name="members">The cluster members, in any order.</param>
+    /// <param name="role">
+    /// The role to scope the selection to, or <c>null</c> to consider every Up member.
+    /// </param>
+    /// <returns>The oldest eligible member, or <c>null</c> when none is Up.</returns>
+    public static Member? OldestUpMember(IEnumerable<Member> members, string? role = null)
+    {
+        ArgumentNullException.ThrowIfNull(members);
+
+        Member? oldest = null;
+        foreach (var member in members)
+        {
+            if (member.Status != MemberStatus.Up) continue;
+            if (role is not null && !member.HasRole(role)) continue;
+            if (oldest is null || member.IsOlderThan(oldest)) oldest = member;
+        }
+
+        return oldest;
+    }
+
+    /// <summary>
+    /// The oldest <see cref="MemberStatus.Up"/> member of <paramref name="cluster"/>, optionally
+    /// scoped to <paramref name="role"/>.
+    /// </summary>
+    /// <param name="cluster">The Akka cluster to evaluate.</param>
+    /// <param name="role">
+    /// The role to scope the selection to, or <c>null</c> to consider every Up member.
+    /// </param>
+    /// <returns>The oldest eligible member, or <c>null</c> when none is Up.</returns>
+    public static Member? OldestUpMember(Cluster cluster, string? role = null)
+    {
+        ArgumentNullException.ThrowIfNull(cluster);
+        return OldestUpMember(cluster.State.Members, role);
+    }
+
+    /// <summary>
+    /// Whether the local node is the active node — Up, and the oldest Up member in the role scope.
+    /// </summary>
+    /// <param name="cluster">The Akka cluster to evaluate.</param>
+    /// <param name="role">
+    /// The role to scope the selection to, or <c>null</c> to consider every Up member.
+    /// </param>
+    /// <returns>
+    /// <c>true</c> when the local member is Up, carries <paramref name="role"/> when one is given,
+    /// and no other eligible member is older.
+    /// </returns>
+    /// <remarks>
+    /// Identity is compared by <see cref="Member.UniqueAddress"/>, not <c>Address</c>: a node that
+    /// restarted on the same host and port is a different member, and during the overlap the cluster
+    /// briefly holds both incarnations.
+    /// </remarks>
+    public static bool SelfIsActive(Cluster cluster, string? role = null)
+    {
+        ArgumentNullException.ThrowIfNull(cluster);
+
+        var self = cluster.SelfMember;
+        if (self.Status != MemberStatus.Up) return false;
+        if (role is not null && !self.HasRole(role)) return false;
+
+        var oldest = OldestUpMember(cluster.State.Members, role);
+        return oldest is not null && oldest.UniqueAddress.Equals(self.UniqueAddress);
+    }
+
+    /// <summary>
+    /// The first role in <paramref name="rolePreference"/> that <paramref name="selfRoles"/> carries
+    /// — the role whose oldest Up member owns this node's active work.
+    /// </summary>
+    /// <param name="selfRoles">The local member's cluster roles.</param>
+    /// <param name="rolePreference">
+    /// Candidate roles in descending precedence, e.g. <c>["admin", "driver"]</c> for a node that
+    /// answers for <c>admin</c> when it is fused and for <c>driver</c> when it is a driver-only node.
+    /// </param>
+    /// <returns>
+    /// The first carried role, or <c>null</c> when the node carries none of them.
+    /// </returns>
+    /// <remarks>
+    /// Precedence matters on a fused node: it must answer for the role its singletons — and the
+    /// traffic being routed to it — are pinned to. Answering for a lower-precedence role could name a
+    /// different node the moment a single-role member joins the mesh.
+    /// </remarks>
+    public static string? ResolveActiveRole(IEnumerable<string> selfRoles, IReadOnlyList<string> rolePreference)
+    {
+        ArgumentNullException.ThrowIfNull(selfRoles);
+        ArgumentNullException.ThrowIfNull(rolePreference);
+
+        if (rolePreference.Count == 0) return null;
+
+        var roles = selfRoles as IReadOnlyCollection<string> ?? selfRoles.ToList();
+        foreach (var candidate in rolePreference)
+        {
+            if (roles.Contains(candidate)) return candidate;
+        }
+
+        return null;
+    }
+}
@@ -1,4 +1,5 @@
 using System.Text.Json;
+using System.Text.Json.Serialization;
 using Microsoft.AspNetCore.Http;
 using Microsoft.Extensions.Diagnostics.HealthChecks;

@@ -16,15 +17,18 @@ namespace ZB.MOM.WW.Health;
 ///   "status": "Healthy|Degraded|Unhealthy",
 ///   "totalDurationMs": 12.34,
 ///   "entries": {
-///     "&lt;name&gt;": { "status": "...", "description": "...", "durationMs": 1.23 }
+///     "&lt;name&gt;": { "status": "...", "description": "...", "durationMs": 1.23, "data": { ... } }
 ///   }
 /// }
 /// </code>
 /// The <c>description</c> key is always present; when a check supplies no description it is emitted
 /// as JSON <c>null</c> (not omitted), matching the spec example and the <c>HealthChecks.UI.Client</c>
-/// shape. The HTTP status code is left to the ASP.NET Core health-checks middleware (Healthy/Degraded
-/// → 200, Unhealthy → 503); this writer only renders the body and sets
-/// <c>Content-Type: application/json</c>.
+/// shape. The <c>data</c> key is the opposite: it is emitted only when the check supplied a non-empty
+/// <see cref="HealthCheckResult.Data"/> dictionary, so payloads from checks that publish no data stay
+/// byte-identical to the pre-0.2.0 shape. Its keys are written verbatim (the camelCase naming policy
+/// applies to the envelope properties, not to dictionary keys). The HTTP status code is left to the
+/// ASP.NET Core health-checks middleware (Healthy/Degraded → 200, Unhealthy → 503); this writer only
+/// renders the body and sets <c>Content-Type: application/json</c>.
 /// </remarks>
 public static class ZbHealthWriter
 {
@@ -60,6 +64,10 @@ public static class ZbHealthWriter
                    Status = e.Value.Status.ToString(),
                    Description = e.Value.Description,
                    DurationMs = e.Value.Duration.TotalMilliseconds,
+                    // HealthReportEntry.Data is never null (empty when the check set none); mapping
+                    // empty → null is what drops the key entirely, keeping data-less payloads
+                    // byte-identical to the pre-0.2.0 shape.
+                    Data = e.Value.Data.Count > 0 ? e.Value.Data : null,
                }),
        };

@@ -78,5 +86,11 @@ public static class ZbHealthWriter
        public string Status { get; init; } = string.Empty;
        public string? Description { get; init; }
        public double DurationMs { get; init; }
+
+        // Per-property ignore, deliberately NOT a global DefaultIgnoreCondition: the writer must keep
+        // emitting "description": null (see SerializerOptions above) while omitting "data" entirely
+        // when a check publishes none.
+        [JsonIgnore(Condition = JsonIgnoreCondition.WhenWritingNull)]
+        public IReadOnlyDictionary<string, object>? Data { get; init; }
    }
 }
@@ -1,137 +0,0 @@
-using Akka.Actor;
-using Microsoft.Extensions.DependencyInjection;
-using Microsoft.Extensions.Diagnostics.HealthChecks;
-using ZB.MOM.WW.Health.Akka;
-
-namespace ZB.MOM.WW.Health.Akka.Tests;
-
-/// <summary>
-/// Table-driven tests for the pure <see cref="ActiveNodeDecision.Evaluate"/> helper covering both
-/// the role-less (ScadaBridge ActiveNode) and role-filtered (OtOpcUa AdminRoleLeader) matrices,
-/// plus the startup-safety null-guards on <see cref="ActiveNodeHealthCheck"/> and
-/// <see cref="AkkaActiveNodeGate"/> when no <see cref="ActorSystem"/> is registered.
-/// </summary>
-public sealed class ActiveNodeDecisionTests
-{
-    // Role-less: requiredRole == null. hasRole is irrelevant. Healthy iff (selfUp && isLeader), else Unhealthy.
-    public static IEnumerable<object?[]> RoleLessCases() => new[]
-    {
-        new object?[] { true, true, false, (string?)null, HealthStatus.Healthy },
-        new object?[] { true, false, false, (string?)null, HealthStatus.Unhealthy },
-        new object?[] { false, true, false, (string?)null, HealthStatus.Unhealthy },
-        new object?[] { false, false, false, (string?)null, HealthStatus.Unhealthy },
-    };
-
-    [Theory]
-    [MemberData(nameof(RoleLessCases))]
-    public void Evaluate_RoleLess(bool selfUp, bool isLeader, bool hasRole, string? requiredRole, HealthStatus expected)
-    {
-        Assert.Equal(expected, ActiveNodeDecision.Evaluate(selfUp, isLeader, hasRole, requiredRole));
-    }
-
-    // Role-filtered: requiredRole != null.
-    //   lacks role               -> Healthy (probe irrelevant for this node)
-    //   has role & is leader     -> Healthy (selfUp is ignored — role-filtered mode only cares about leadership)
-    //   has role & not leader    -> Degraded
-    public static IEnumerable<object[]> RoleFilteredCases() => new[]
-    {
-        // node lacks the role -> Healthy regardless of selfUp / isLeader
-        new object[] { true, true, false, "admin", HealthStatus.Healthy },
-        new object[] { true, false, false, "admin", HealthStatus.Healthy },
-        new object[] { false, false, false, "admin", HealthStatus.Healthy },
-        // node carries the role and is leader -> Healthy (selfUp=true)
-        new object[] { true, true, true, "admin", HealthStatus.Healthy },
-        // node carries the role and is leader -> Healthy (selfUp=false: role-filtered mode ignores selfUp)
-        new object[] { false, true, true, "admin", HealthStatus.Healthy },
-        // node carries the role but is not leader -> Degraded
-        new object[] { true, false, true, "admin", HealthStatus.Degraded },
-        new object[] { false, false, true, "admin", HealthStatus.Degraded },
-    };
-
-    [Theory]
-    [MemberData(nameof(RoleFilteredCases))]
-    public void Evaluate_RoleFiltered(bool selfUp, bool isLeader, bool hasRole, string? requiredRole, HealthStatus expected)
-    {
-        Assert.Equal(expected, ActiveNodeDecision.Evaluate(selfUp, isLeader, hasRole, requiredRole));
-    }
-
-    [Fact]
-    public async Task HealthCheck_RoleLess_NoActorSystem_ReturnsDegraded()
-    {
-        var provider = new ServiceCollection().BuildServiceProvider();
-        var check = new ActiveNodeHealthCheck(provider);
-
-        var result = await check.CheckHealthAsync(NewContext(check));
-
-        Assert.Equal(HealthStatus.Degraded, result.Status);
-    }
-
-    [Fact]
-    public async Task HealthCheck_RoleFiltered_NoActorSystem_ReturnsDegraded()
-    {
-        var provider = new ServiceCollection().BuildServiceProvider();
-        var check = new ActiveNodeHealthCheck(provider, "admin");
-
-        var result = await check.CheckHealthAsync(NewContext(check));
-
-        Assert.Equal(HealthStatus.Degraded, result.Status);
-    }
-
-    [Fact]
-    public void Gate_NoActorSystem_IsActiveNodeFalse()
-    {
-        var provider = new ServiceCollection().BuildServiceProvider();
-        var gate = new AkkaActiveNodeGate(provider);
-
-        Assert.False(gate.IsActiveNode);
-    }
-
-    [Fact]
-    public async Task HealthCheck_RoleLess_ClusterInaccessible_ReturnsDegraded()
-    {
-        // ActorSystem present but Akka.Cluster not configured → Cluster.Get throws. The check must
-        // return Degraded (startup-safety rule), not let the exception escape (→ Unhealthy).
-        using var system = ActorSystem.Create("plain-no-cluster-roleless");
-        try
-        {
-            var provider = new ServiceCollection()
-                .AddSingleton(system)
-                .BuildServiceProvider();
-            var check = new ActiveNodeHealthCheck(provider);
-
-            var result = await check.CheckHealthAsync(NewContext(check));
-
-            Assert.Equal(HealthStatus.Degraded, result.Status);
-        }
-        finally
-        {
-            await system.Terminate();
-        }
-    }
-
-    [Fact]
-    public async Task HealthCheck_RoleFiltered_ClusterInaccessible_ReturnsDegraded()
-    {
-        using var system = ActorSystem.Create("plain-no-cluster-rolefiltered");
-        try
-        {
-            var provider = new ServiceCollection()
-                .AddSingleton(system)
-                .BuildServiceProvider();
-            var check = new ActiveNodeHealthCheck(provider, "admin");
-
-            var result = await check.CheckHealthAsync(NewContext(check));
-
-            Assert.Equal(HealthStatus.Degraded, result.Status);
-        }
-        finally
-        {
-            await system.Terminate();
-        }
-    }
-
-    private static HealthCheckContext NewContext(IHealthCheck check) => new()
-    {
-        Registration = new HealthCheckRegistration("active-node", check, HealthStatus.Unhealthy, tags: null),
-    };
-}
@@ -0,0 +1,113 @@
+using Akka.Actor;
+using Akka.Configuration;
+using Microsoft.Extensions.DependencyInjection;
+using Microsoft.Extensions.Diagnostics.HealthChecks;
+using ZB.MOM.WW.Health.Akka;
+
+namespace ZB.MOM.WW.Health.Akka.Tests;
+
+/// <summary>
+/// The states a node passes through on the way up, where the active tier must not report a booting
+/// node as a failed standby — now that the tier is a real 503, an Unhealthy here would pull the node
+/// out of an orchestrator's pool while it starts.
+/// </summary>
+public sealed class ActiveNodeStartupSafetyTests
+{
+    [Fact]
+    public async Task NoActorSystem_IsDegraded()
+    {
+        var check = new ActiveNodeHealthCheck(new ServiceCollection().BuildServiceProvider());
+
+        var result = await check.CheckHealthAsync(NewContext(check));
+
+        Assert.Equal(HealthStatus.Degraded, result.Status);
+        // Description-only: an empty data dictionary is what makes the writer omit "data" entirely.
+        Assert.Empty(result.Data);
+    }
+
+    [Fact]
+    public async Task NoActorSystem_RoleScoped_IsDegraded()
+    {
+        var check = new ActiveNodeHealthCheck(new ServiceCollection().BuildServiceProvider(), "admin");
+
+        Assert.Equal(HealthStatus.Degraded, (await check.CheckHealthAsync(NewContext(check))).Status);
+    }
+
+    [Fact]
+    public void Gate_NoActorSystem_IsNotActive()
+    {
+        Assert.False(new AkkaActiveNodeGate(new ServiceCollection().BuildServiceProvider()).IsActiveNode);
+    }
+
+    [Fact]
+    public async Task ClusterNotConfigured_IsDegraded()
+    {
+        // ActorSystem present but Akka.Cluster not configured → Cluster.Get throws. The check must
+        // return Degraded (the spec's startup-safety rule), not let the exception escape (→ Unhealthy).
+        using var system = ActorSystem.Create("active-node-no-cluster");
+        try
+        {
+            var provider = new ServiceCollection().AddSingleton(system).BuildServiceProvider();
+            var check = new ActiveNodeHealthCheck(provider);
+
+            Assert.Equal(HealthStatus.Degraded, (await check.CheckHealthAsync(NewContext(check))).Status);
+        }
+        finally
+        {
+            await system.Terminate();
+        }
+    }
+
+    [Fact]
+    public async Task Gate_ClusterNotConfigured_IsNotActive()
+    {
+        using var system = ActorSystem.Create("active-node-gate-no-cluster");
+        try
+        {
+            var provider = new ServiceCollection().AddSingleton(system).BuildServiceProvider();
+
+            Assert.False(new AkkaActiveNodeGate(provider).IsActiveNode);
+        }
+        finally
+        {
+            await system.Terminate();
+        }
+    }
+
+    [Fact]
+    public async Task ClusteredButNeverJoined_IsUnhealthy_NobodyIsActive()
+    {
+        // A cold start: the cluster provider is up but no member has reached Up. Nobody is active, so
+        // nobody may claim to be — this is the one case that must NOT be Degraded, because a cluster
+        // that never forms would otherwise sit at 200 forever.
+        var config = ConfigurationFactory.ParseString("""
+            akka {
+              loglevel = "WARNING"
+              actor.provider = cluster
+              remote.dot-netty.tcp { hostname = "127.0.0.1", port = 0 }
+              cluster.seed-nodes = []
+            }
+            """);
+        var system = ActorSystem.Create("active-node-never-joined", config);
+        try
+        {
+            var provider = new ServiceCollection().AddSingleton(system).BuildServiceProvider();
+            var check = new ActiveNodeHealthCheck(provider);
+
+            var result = await check.CheckHealthAsync(NewContext(check));
+
+            Assert.Equal(HealthStatus.Unhealthy, result.Status);
+            Assert.Contains("No Up member", result.Description);
+            Assert.False(new AkkaActiveNodeGate(provider).IsActiveNode);
+        }
+        finally
+        {
+            await system.Terminate();
+        }
+    }
+
+    private static HealthCheckContext NewContext(IHealthCheck check) => new()
+    {
+        Registration = new HealthCheckRegistration("active-node", check, HealthStatus.Unhealthy, tags: null),
+    };
+}
@@ -0,0 +1,56 @@
+using ZB.MOM.WW.Health.Akka;
+
+namespace ZB.MOM.WW.Health.Akka.Tests;
+
+/// <summary>
+/// Covers <see cref="ClusterActiveNode.ResolveActiveRole"/> — which role a node competes for. Pure,
+/// so no cluster is needed; the age-ordering half of the rule is pinned in
+/// <see cref="ClusterActiveNodeTests"/> against a real two-node cluster.
+/// </summary>
+public sealed class ActiveRoleResolutionTests
+{
+    [Fact]
+    public void EmptyPreference_IsUnscoped()
+    {
+        // Unscoped: every Up member competes for one active slot (the ScadaBridge central pattern).
+        Assert.Null(ClusterActiveNode.ResolveActiveRole(["admin", "driver"], []));
+    }
+
+    [Fact]
+    public void FusedNode_AnswersForTheHighestPrecedenceRoleItCarries()
+    {
+        Assert.Equal(
+            "admin",
+            ClusterActiveNode.ResolveActiveRole(["admin", "driver", "cluster-MAIN"], ["admin", "driver"]));
+    }
+
+    [Fact]
+    public void SingleRoleNode_AnswersForThatRole()
+    {
+        // The regression the OtOpcUa fix existed for: under the old role-filtered check every
+        // driver-only node reported Healthy — "or not a role member" — so all of them called
+        // themselves active and no consumer could find the active node of a site cluster. lmxopcua#494.
+        Assert.Equal(
+            "driver",
+            ClusterActiveNode.ResolveActiveRole(["driver", "cluster-SITE-A"], ["admin", "driver"]));
+    }
+
+    [Fact]
+    public void PreferenceOrderWins_NotTheOrderTheNodeListsItsRoles()
+    {
+        // Self-role order is incidental (it comes from config parsing); precedence must come from the
+        // caller's list alone, or a fused node's answer would depend on how its roles were spelled.
+        Assert.Equal(
+            "admin",
+            ClusterActiveNode.ResolveActiveRole(["driver", "admin"], ["admin", "driver"]));
+    }
+
+    [Fact]
+    public void NodeCarryingNoneOfThePreferredRoles_ResolvesToNull()
+    {
+        // Reachable, not merely defensive: OtOpcUa's RoleParser admits a dev-only or cluster-role-only
+        // node. Such a node owns no active work — the health check maps this to Unhealthy so it stays
+        // out of the active pool.
+        Assert.Null(ClusterActiveNode.ResolveActiveRole(["dev", "cluster-MAIN"], ["admin", "driver"]));
+    }
+}
@@ -0,0 +1,107 @@
+using Akka.Actor;
+using Akka.Cluster;
+using Akka.TestKit.Xunit2;
+using Microsoft.Extensions.DependencyInjection;
+using Microsoft.Extensions.Diagnostics.HealthChecks;
+using ZB.MOM.WW.Health.Akka;
+
+namespace ZB.MOM.WW.Health.Akka.Tests;
+
+/// <summary>
+/// Covers the cluster-view <c>data</c> dictionary that <see cref="AkkaClusterHealthCheck"/> publishes
+/// from 0.2.0 on (consumed by the family overview dashboard to show each cluster's current leader).
+/// Uses a real single-node self-joined cluster via <see cref="TestKit"/> — the values come off
+/// <see cref="Cluster"/>, which cannot be faked, so a formed cluster is the only honest fixture.
+/// </summary>
+public sealed class AkkaClusterDataTests : TestKit
+{
+    private const string ClusterConfig = """
+        akka {
+          actor.provider = cluster
+          remote.dot-netty.tcp {
+            hostname = "127.0.0.1"
+            port = 0
+          }
+          cluster {
+            roles = ["driver", "admin"]
+            # Keep the single-node self-join deterministic and fast in-test.
+            min-nr-of-members = 1
+          }
+        }
+        """;
+
+    private static readonly TimeSpan FormationTimeout = TimeSpan.FromSeconds(30);
+
+    public AkkaClusterDataTests()
+        : base(ClusterConfig, nameof(AkkaClusterDataTests))
+    {
+    }
+
+    /// <summary>Self-joins the test node and waits until it is Up with a converged leader.</summary>
+    private Cluster FormSingleNodeCluster()
+    {
+        var cluster = Cluster.Get(Sys);
+        cluster.Join(cluster.SelfAddress);
+        AwaitCondition(
+            () => cluster.SelfMember.Status == MemberStatus.Up && cluster.State.Leader is not null,
+            FormationTimeout);
+        return cluster;
+    }
+
+    [Fact]
+    public void BuildClusterData_SingleNodeCluster_ReportsSelfAsLeader()
+    {
+        var cluster = FormSingleNodeCluster();
+
+        var data = AkkaClusterHealthCheck.BuildClusterData(cluster);
+
+        Assert.Equal(cluster.SelfAddress.ToString(), Assert.IsType<string>(data["leader"]));
+        Assert.Equal(cluster.SelfAddress.ToString(), Assert.IsType<string>(data["selfAddress"]));
+        Assert.Equal(1, Assert.IsType<int>(data["memberCount"]));
+        Assert.Equal(0, Assert.IsType<int>(data["unreachableCount"]));
+    }
+
+    [Fact]
+    public void BuildClusterData_RolesRoundTripSorted()
+    {
+        var cluster = FormSingleNodeCluster();
+
+        var roles = Assert.IsType<string[]>(AkkaClusterHealthCheck.BuildClusterData(cluster)["selfRoles"]);
+
+        // Sorted, so the payload is stable across restarts; the configured roles survive verbatim.
+        Assert.Equal(new[] { "admin", "driver" }, roles);
+    }
+
+    [Fact]
+    public async Task CheckHealth_FormedCluster_HealthyResultCarriesClusterData()
+    {
+        var cluster = FormSingleNodeCluster();
+        var provider = new ServiceCollection().AddSingleton(Sys).BuildServiceProvider();
+        var check = new AkkaClusterHealthCheck(provider, AkkaClusterStatusPolicy.Default);
+
+        var result = await check.CheckHealthAsync(NewContext(check));
+
+        Assert.Equal(HealthStatus.Healthy, result.Status);
+        Assert.Equal(cluster.SelfAddress.ToString(), result.Data["leader"]);
+        Assert.Equal(1, result.Data["memberCount"]);
+    }
+
+    [Fact]
+    public async Task CheckHealth_NoActorSystem_CarriesNoData()
+    {
+        // The startup-safety paths stay description-only: an empty data dictionary is what makes the
+        // writer omit the "data" key entirely, so a not-yet-started node's payload is unchanged.
+        var provider = new ServiceCollection().BuildServiceProvider();
+        var check = new AkkaClusterHealthCheck(provider, AkkaClusterStatusPolicy.Default);
+
+        var result = await check.CheckHealthAsync(NewContext(check));
+
+        Assert.Equal(HealthStatus.Degraded, result.Status);
+        Assert.Empty(result.Data);
+    }
+
+    private static HealthCheckContext NewContext(IHealthCheck check) => new()
+    {
+        Registration = new HealthCheckRegistration("akka-cluster", check, HealthStatus.Unhealthy, tags: null),
+    };
+}
@@ -88,6 +88,8 @@ public sealed class AkkaClusterStatusPolicyTests
            var result = await check.CheckHealthAsync(NewContext(check));

            Assert.Equal(HealthStatus.Degraded, result.Status);
+            // No cluster to describe → no data, so the emitted payload stays the pre-0.2.0 shape.
+            Assert.Empty(result.Data);
        }
        finally
        {
@@ -0,0 +1,237 @@
+using Akka.Actor;
+using Akka.Cluster;
+using Akka.Configuration;
+using Microsoft.Extensions.DependencyInjection;
+using Microsoft.Extensions.Diagnostics.HealthChecks;
+using ZB.MOM.WW.Health.Akka;
+
+namespace ZB.MOM.WW.Health.Akka.Tests;
+
+/// <summary>
+/// Pins <b>which</b> node the active tier names, against a real two-node cluster deliberately built
+/// so the oldest member is <i>not</i> the lowest-addressed one.
+/// </summary>
+/// <remarks>
+///     <para>
+///         The divergence is the whole point. Akka offers two different "first" members:
+///         <c>ClusterState.Leader</c> / <c>RoleLeader</c> is the <b>lowest-addressed</b> Up member
+///         (host, then port — nothing to do with time), while <b>oldest</b> is the member with the
+///         lowest up-number, which is what <c>ClusterSingletonManager</c> uses to place singletons.
+///     </para>
+///     <para>
+///         They coincide on a freshly-formed cluster, which is why every single-node and happy-path
+///         test missed it and why the leader-based check survived to 0.2.1. They diverge the moment a
+///         node restarts: the restarted node rejoins as the youngest while keeping its address, so if
+///         it holds the lower address it becomes leader while the singletons — and every piece of work
+///         they own — stay on the other node. This fixture reproduces that state directly instead of
+///         waiting for a restart.
+///     </para>
+/// </remarks>
+public sealed class ClusterActiveNodeTests : IAsyncLifetime
+{
+    // Fixed, unusual ports: the test needs a deterministic address ordering, which port 0 cannot give.
+    // Distinct from OtOpcUa's equivalent fixture (19_530/19_531) so the two suites can run at once.
+    private const int OldestPort = 19_541;   // joins FIRST  -> oldest, but the HIGHER address
+    private const int YoungestPort = 19_540; // joins SECOND -> cluster leader, the LOWER address
+
+    private const string SystemName = "health-active-node";
+
+    private ActorSystem? _oldest;
+    private ActorSystem? _youngest;
+
+    private static Config NodeConfig(int port, string roles) => ConfigurationFactory.ParseString($$"""
+        akka {
+            loglevel = "WARNING"
+            actor.provider = "Akka.Cluster.ClusterActorRefProvider, Akka.Cluster"
+            remote.dot-netty.tcp {
+                hostname = "127.0.0.1"
+                port = {{port}}
+            }
+            cluster {
+                seed-nodes = ["akka.tcp://{{SystemName}}@127.0.0.1:{{OldestPort}}"]
+                roles = [{{roles}}]
+                min-nr-of-members = 1
+                run-coordinated-shutdown-when-down = off
+                downing-provider-class = ""
+            }
+        }
+        """);
+
+    public async Task InitializeAsync()
+    {
+        // Order matters: the seed forms the cluster and is therefore the oldest member. The oldest
+        // node is fused (admin + driver); the younger one is driver-only, so the role-scoped
+        // assertions below can distinguish "oldest of this role" from "oldest overall".
+        _oldest = ActorSystem.Create(SystemName, NodeConfig(OldestPort, "\"admin\", \"driver\""));
+        await WaitForUpAsync(_oldest, expectedMembers: 1);
+
+        _youngest = ActorSystem.Create(SystemName, NodeConfig(YoungestPort, "\"driver\""));
+        await WaitForUpAsync(_oldest, expectedMembers: 2);
+        await WaitForUpAsync(_youngest, expectedMembers: 2);
+    }
+
+    public async Task DisposeAsync()
+    {
+        if (_youngest is not null) await _youngest.Terminate();
+        if (_oldest is not null) await _oldest.Terminate();
+    }
+
+    private static async Task WaitForUpAsync(ActorSystem system, int expectedMembers)
+    {
+        var cluster = Cluster.Get(system);
+        var deadline = DateTime.UtcNow + TimeSpan.FromSeconds(30);
+        while (DateTime.UtcNow < deadline)
+        {
+            if (cluster.State.Members.Count(m => m.Status == MemberStatus.Up) >= expectedMembers) return;
+            await Task.Delay(100);
+        }
+
+        throw new TimeoutException(
+            $"cluster on port {cluster.SelfAddress.Port} never saw {expectedMembers} Up members "
+            + $"(saw {cluster.State.Members.Count(m => m.Status == MemberStatus.Up)})");
+    }
+
+    /// <summary>
+    /// Sanity check on the fixture itself: if the two orderings did not actually diverge, every
+    /// assertion below would pass for the wrong reason.
+    /// </summary>
+    [Fact]
+    public void Fixture_actually_produces_divergent_age_and_address_ordering()
+    {
+        var state = Cluster.Get(_oldest!).State;
+
+        Assert.Equal(YoungestPort, state.Leader!.Port);
+        Assert.Equal(YoungestPort, state.RoleLeader("driver")!.Port);
+    }
+
+    [Fact]
+    public void OldestUpMember_IsTheOldest_NotTheLeader()
+    {
+        var members = Cluster.Get(_oldest!).State.Members;
+
+        Assert.Equal(OldestPort, ClusterActiveNode.OldestUpMember(members)!.Address.Port);
+        Assert.Equal(OldestPort, ClusterActiveNode.OldestUpMember(members, "driver")!.Address.Port);
+    }
+
+    [Fact]
+    public void OldestUpMember_IsRoleScoped()
+    {
+        var members = Cluster.Get(_oldest!).State.Members;
+
+        // Only the fused node carries admin, so it owns the admin slot even though it is not the
+        // lowest-addressed member.
+        Assert.Equal(OldestPort, ClusterActiveNode.OldestUpMember(members, "admin")!.Address.Port);
+
+        // A role nobody carries has no owner.
+        Assert.Null(ClusterActiveNode.OldestUpMember(members, "no-such-role"));
+    }
+
+    [Fact]
+    public void SelfIsActive_IsTrueOnlyOnTheOldestNode()
+    {
+        Assert.True(ClusterActiveNode.SelfIsActive(Cluster.Get(_oldest!)));
+        Assert.False(ClusterActiveNode.SelfIsActive(Cluster.Get(_youngest!)));
+    }
+
+    [Fact]
+    public void SelfIsActive_RoleScoped_YoungestOwnsNoSlot()
+    {
+        // Driver-scoped: the fused node is still the oldest driver, so the driver-only node is standby.
+        Assert.True(ClusterActiveNode.SelfIsActive(Cluster.Get(_oldest!), "driver"));
+        Assert.False(ClusterActiveNode.SelfIsActive(Cluster.Get(_youngest!), "driver"));
+
+        // Admin-scoped: the driver-only node does not carry admin, so it can never be the admin active
+        // node — even though it is the cluster leader.
+        Assert.False(ClusterActiveNode.SelfIsActive(Cluster.Get(_youngest!), "admin"));
+    }
+
+    [Fact]
+    public async Task HealthCheck_IsHealthyOnTheOldestAndUnhealthyOnTheLeader()
+    {
+        var onOldest = await CheckAsync(_oldest!, new ActiveNodeHealthCheckOptions());
+        var onYoungest = await CheckAsync(_youngest!, new ActiveNodeHealthCheckOptions());
+
+        Assert.Equal(HealthStatus.Healthy, onOldest.Status);
+
+        // The node the old leader-based check would have called active. Exactly one 200 per cluster is
+        // the property an orchestrator's active-tier routing depends on.
+        Assert.Equal(HealthStatus.Unhealthy, onYoungest.Status);
+    }
+
+    [Fact]
+    public async Task HealthCheck_DataNamesTheActiveNode_EvenFromAStandby()
+    {
+        // A standby reporting *who* is active is what lets a dashboard render the pair from either
+        // node's payload alone (the 0.2.0 per-entry data object).
+        var result = await CheckAsync(_youngest!, new ActiveNodeHealthCheckOptions());
+
+        Assert.Equal($"akka.tcp://{SystemName}@127.0.0.1:{OldestPort}", result.Data["activeNode"]);
+        Assert.Equal($"akka.tcp://{SystemName}@127.0.0.1:{YoungestPort}", result.Data["selfAddress"]);
+    }
+
+    [Fact]
+    public async Task HealthCheck_RolePreference_FusedNodeAnswersForAdmin()
+    {
+        var options = new ActiveNodeHealthCheckOptions { RolePreference = ["admin", "driver"] };
+
+        var onOldest = await CheckAsync(_oldest!, options);
+        var onYoungest = await CheckAsync(_youngest!, options);
+
+        Assert.Equal(HealthStatus.Healthy, onOldest.Status);
+        Assert.Equal("admin", onOldest.Data["activeRole"]);
+
+        // Resolves to driver (it has no admin role) and loses to the older fused node.
+        Assert.Equal(HealthStatus.Unhealthy, onYoungest.Status);
+        Assert.Equal("driver", onYoungest.Data["activeRole"]);
+    }
+
+    [Fact]
+    public async Task HealthCheck_NodeOwningNoneOfThePreferredRoles_IsUnhealthyByDefault()
+    {
+        var options = new ActiveNodeHealthCheckOptions { RolePreference = ["no-such-role"] };
+
+        var result = await CheckAsync(_oldest!, options);
+
+        // Not "not applicable ⇒ Healthy": a node that owns no active work must stay out of the pool.
+        Assert.Equal(HealthStatus.Unhealthy, result.Status);
+        Assert.Contains("carries none of the active roles", result.Description);
+        Assert.DoesNotContain("activeRole", result.Data.Keys);
+    }
+
+    [Fact]
+    public async Task HealthCheck_NoActiveRoleStatus_IsOverridable()
+    {
+        var options = new ActiveNodeHealthCheckOptions
+        {
+            RolePreference = ["no-such-role"],
+            NoActiveRoleStatus = HealthStatus.Healthy,
+        };
+
+        Assert.Equal(HealthStatus.Healthy, (await CheckAsync(_oldest!, options)).Status);
+    }
+
+    [Fact]
+    public void Gate_IsActiveOnlyOnTheOldestNode()
+    {
+        Assert.True(new AkkaActiveNodeGate(ProviderFor(_oldest!)).IsActiveNode);
+        Assert.False(new AkkaActiveNodeGate(ProviderFor(_youngest!)).IsActiveNode);
+
+        // Role-scoped: the driver-only node cannot gate admin work.
+        Assert.False(new AkkaActiveNodeGate(ProviderFor(_youngest!), "admin").IsActiveNode);
+    }
+
+    private static IServiceProvider ProviderFor(ActorSystem system) =>
+        new ServiceCollection().AddSingleton(system).BuildServiceProvider();
+
+    private static async Task<HealthCheckResult> CheckAsync(
+        ActorSystem system,
+        ActiveNodeHealthCheckOptions options)
+    {
+        var check = new ActiveNodeHealthCheck(ProviderFor(system), options);
+        var context = new HealthCheckContext
+        {
+            Registration = new HealthCheckRegistration("active-node", check, HealthStatus.Unhealthy, tags: null),
+        };
+        return await check.CheckHealthAsync(context);
+    }
+}
@@ -5,6 +5,7 @@
  </PropertyGroup>

  <ItemGroup>
+    <PackageReference Include="Akka.TestKit.Xunit2" />
    <PackageReference Include="coverlet.collector" />
    <PackageReference Include="Microsoft.NET.Test.Sdk" />
    <PackageReference Include="xunit" />
@@ -22,8 +22,11 @@ public sealed class ResponseWriterTests
    {
        private readonly HealthCheckResult _result;

-        public StubHealthCheck(HealthStatus status, string? description = null) =>
-            _result = new HealthCheckResult(status, description);
+        public StubHealthCheck(
+            HealthStatus status,
+            string? description = null,
+            IReadOnlyDictionary<string, object>? data = null) =>
+            _result = new HealthCheckResult(status, description, exception: null, data: data);

        public Task<HealthCheckResult> CheckHealthAsync(
            HealthCheckContext context,
@@ -31,12 +34,14 @@ public sealed class ResponseWriterTests
    }

    private static async Task<HttpResponseMessage> GetReadyAsync(
-        HealthStatus status, string? description = "db reachable")
+        HealthStatus status,
+        string? description = "db reachable",
+        IReadOnlyDictionary<string, object>? data = null)
    {
        var builder = WebApplication.CreateBuilder();
        builder.WebHost.UseTestServer();
        builder.Services.AddHealthChecks()
-            .AddCheck("db", new StubHealthCheck(status, description), tags: new[] { ZbHealthTags.Ready });
+            .AddCheck("db", new StubHealthCheck(status, description, data), tags: new[] { ZbHealthTags.Ready });

        await using var app = builder.Build();
        app.MapZbHealth();
@@ -84,6 +89,51 @@ public sealed class ResponseWriterTests
        Assert.Equal(JsonValueKind.Null, description.ValueKind);
    }

+    [Fact]
+    public async Task ReadyEndpoint_EntryWithData_EmitsDataObject_WithVerbatimKeys()
+    {
+        // The envelope properties are camelCased by the naming policy, but a check's own data keys are
+        // dictionary keys and must survive verbatim — the dashboard reads them by exact name
+        // (entries["akka-cluster"].data.leader), so a naming policy applied here would silently break
+        // the contract. Values may be any JSON-friendly type (string / int / string[]).
+        var data = new Dictionary<string, object>
+        {
+            ["leader"] = "akka.tcp://sys@node-1:4053",
+            ["memberCount"] = 2,
+            ["selfRoles"] = new[] { "driver", "site-a" },
+        };
+
+        var response = await GetReadyAsync(HealthStatus.Healthy, data: data);
+
+        Assert.Equal(HttpStatusCode.OK, response.StatusCode);
+
+        using var doc = JsonDocument.Parse(await response.Content.ReadAsStringAsync());
+        var entryData = doc.RootElement.GetProperty("entries").GetProperty("db").GetProperty("data");
+
+        Assert.Equal("akka.tcp://sys@node-1:4053", entryData.GetProperty("leader").GetString());
+        Assert.Equal(2, entryData.GetProperty("memberCount").GetInt32());
+        Assert.Equal(
+            new[] { "driver", "site-a" },
+            entryData.GetProperty("selfRoles").EnumerateArray().Select(static e => e.GetString()).ToArray());
+    }
+
+    [Fact]
+    public async Task ReadyEndpoint_EntryWithoutData_OmitsDataKeyEntirely()
+    {
+        // The 0.2.0 non-breaking invariant: a check that publishes no data must produce exactly the
+        // pre-0.2.0 body. Asserted on the raw JSON string so a `"data": null` regression (which
+        // TryGetProperty would still accept as "present") fails here.
+        var response = await GetReadyAsync(HealthStatus.Healthy);
+
+        var body = await response.Content.ReadAsStringAsync();
+
+        Assert.DoesNotContain("\"data\"", body, StringComparison.Ordinal);
+
+        using var doc = JsonDocument.Parse(body);
+        var db = doc.RootElement.GetProperty("entries").GetProperty("db");
+        Assert.False(db.TryGetProperty("data", out _), "data key must be absent when the check set none");
+    }
+
    [Fact]
    public async Task ReadyEndpoint_Degraded_Returns200_WithDegradedStatus()
    {
@@ -5,7 +5,7 @@
    <Nullable>enable</Nullable>
    <ImplicitUsings>enable</ImplicitUsings>
    <LangVersion>latest</LangVersion>
-    <Version>0.1.3</Version>
+    <Version>0.3.4</Version>
    <ManagePackageVersionsCentrally>true</ManagePackageVersionsCentrally>
    <TreatWarningsAsErrors>true</TreatWarningsAsErrors>
    <!-- NU1510 (package-pruning advice): the Replication and Tests projects carry a
@@ -10,16 +10,143 @@ Change capture uses SQLite triggers writing a single HLC-stamped oplog inside th
 consumer's own write transaction (crash-consistent, non-bypassable). Convergence is
 last-writer-wins over a hybrid logical clock. All timestamps are UTC.

-## Packages (all `0.1.3`)
+## Packages (all `0.3.4`)

 | Package | Contents |
 |---|---|
-| `ZB.MOM.WW.LocalDb` | Embedded DB host: `ILocalDb` SQL surface (`ExecuteAsync` / `QueryAsync` / transactions), `RegisterReplicated(table)` (PK validation + trigger install + oplog schema), `HybridLogicalClock`, `AddZbLocalDb()` DI. |
+| `ZB.MOM.WW.LocalDb` | Embedded DB host: `ILocalDb` SQL surface (`ExecuteAsync` / `QueryAsync` / transactions), `RegisterReplicated(table[, baselineExistingRows])` (PK + BLOB + foreign-key validation, trigger install, oplog schema), `DeregisterReplicated(table)` (trigger removal + oplog/ledger prune), `HybridLogicalClock`, `AddZbLocalDb()` DI. |
 | `ZB.MOM.WW.LocalDb.Contracts` | `localdb_sync.v1` proto + generated gRPC types (packable; a future peer could be non-.NET). |
 | `ZB.MOM.WW.LocalDb.Replication` | gRPC sync engine: passive `LocalDbSyncService`, initiating `SyncBackgroundService`, `MaintenanceBackgroundService`, LWW apply, watermark acks + pruning, snapshot resync, `ISyncStatus`, `LocalDbMetrics`, `AddZbLocalDbReplication()` / `MapZbLocalDbSync()`. |

 > Status: **published to the Gitea feed; adopted by ScadaBridge (Phases 1+2) and OtOpcUa (Phase 1).**
 >
+> - `0.3.4` (2026-08-20) — **a converged pair no longer re-streams a full snapshot on every
+>   reconnect once the sender's oplog has been ack-pruned to empty.** Found on the ScadaBridge
+>   docker rig (its register row 106): a rebuilt node (watermark 0) was healed by a snapshot
+>   stamped `as_of_seq = MAX(seq)` — which is 0 on an empty oplog — so its MAX-guarded watermark
+>   stayed 0, and the healthy side's empty-oplog horizon rule (`0.1.2`: streamable from
+>   `last_acked_seq + 1`) saw a gap on every later handshake and re-sent the whole database
+>   (127k rows per restart on the rig) until it happened to write a new row. The as-of is now
+>   `max(MAX(seq), last_acked_seq)`: every delta pruned below `last_acked_seq` is already folded
+>   into the rows the snapshot carries, so that is the floor it may claim. Sender-side only,
+>   wire-compatible, no schema change; pinned by
+>   `WipedPeer_SnapshotLiftsItsWatermark_SoTheNextReconnectSendsNoSecondSnapshot`.
+> - `0.3.3` (2026-08-20) — two latent apply-path hazards from the ScadaBridge adversarial
+>   architecture review (P4-7). Neither had a live trigger in any current consumer; both are the
+>   kind that surface as silent divergence, not as an error. Wire-compatible: no proto change, no
+>   `lib_schema_version` bump, no on-disk schema change.
+>   - **A replicated table may no longer be the parent of a row-mutating `ON DELETE` foreign key.**
+>     The apply path writes remote rows with `INSERT OR REPLACE`, and REPLACE resolves a PK conflict
+>     by DELETING the existing row first. Connections run `PRAGMA foreign_keys=ON`, so that implicit
+>     delete fires the child's `ON DELETE CASCADE` / `SET NULL` / `SET DEFAULT` — while the capture
+>     guard is raised. The child rows change on the applying node ONLY, produce no oplog entry, and
+>     the pair diverges with nothing able to detect it (LWW compares only rows it was told about).
+>     `RegisterReplicated` now inspects every foreign key in the database and refuses such a table
+>     with a `LocalDbRegistrationException` naming the offending child column and action.
+>     `RESTRICT` / `NO ACTION` remain fine — they abort the statement instead of mutating rows
+>     behind capture's back, which the apply path already handles as per-entry poison. Checked at
+>     registration; SQLite cannot add a foreign key to an existing table, so a child table created
+>     later is caught at the next startup's registration rather than at the DDL.
+>   - **An entry's row write and its ledger write are now atomic with each other.** They are two
+>     statements, and SQLite aborts only the failing one — so a statement-level fault between them
+>     (the case the per-entry poison catch exists for) committed a half-applied entry: a tombstone
+>     whose table row was gone while `__localdb_row_version` still called it live, which the snapshot
+>     streamer can only warn about and skip, so the peer never learned the row died; or an upsert
+>     whose new row carried the OLD version and therefore lost the next comparison it should have
+>     won. The pair is wrapped in a per-entry `SAVEPOINT` that is rolled back before dead-lettering,
+>     so a poison entry leaves both writes or neither and the rest of the batch is untouched.
+> - `0.3.2` (2026-08-19) — two Medium findings from the ScadaBridge adversarial architecture review
+>   (P4-1, P4-3). Wire-compatible with every earlier peer: no proto change, no `lib_schema_version`
+>   bump, no on-disk schema change — the keepalive is transport-level and the rest is local policy.
+>   - **A live session now ends itself when a cap prune eats deltas it has not sent.** The pump reads
+>     `seq > sentThruSeq`, which resumes SILENTLY above a pruned hole; the peer acks the later seqs
+>     and the pair is durably divergent for the whole life of that session. `needs_snapshot` (set in
+>     the prune's own transaction) only heals a RECONNECT, and nothing forced one — so on a stable
+>     LAN with a slow-but-alive peer the divergence was unbounded. Every pump iteration now compares
+>     the oplog horizon (`MIN(seq)`, one b-tree seek) against its sent watermark and throws
+>     `LocalDbOplogGapException` when deltas it still owes are gone. The horizon, not the flag,
+>     because the flag is also raised for reasons that do not invalidate the running session.
+>   - **The sync stream has HTTP/2 keepalive, and the server's single-stream guard is newest-wins.**
+>     Neither end ever timed out on silence (a converged pair is legitimately quiet), so a hard crash
+>     or partition of the dialling node left the passive node's handler parked in `MoveNext` on a
+>     half-open stream — holding the guard, reporting itself connected, and rejecting the rebooted
+>     peer's reconnects with `AlreadyExists`, indefinitely on an idle pair. The initiator now dials
+>     with `KeepAlivePingDelay` / `KeepAlivePingTimeout` (new options, 60s/20s, `Always` policy so
+>     idle connections ping too), and a fresh authenticated `Sync` cancels the incumbent and waits
+>     for it to unwind before starting — the guard serializes sessions, it was never a fence, and an
+>     authenticated same-key dialler is by definition the peer. An incumbent that will not yield
+>     within the takeover timeout still gets `AlreadyExists`, because two live sessions over one
+>     database is worse than a refused dial.
+> - `0.3.1` (2026-08-18) — a faulting session flushes its control lane before cancelling its writer,
+>   so a locally-detected handshake mismatch actually reaches the peer instead of leaving it to
+>   reconnect-loop against a generic disconnect.
+> - `0.3.0` (2026-08-15) — **the snapshot-resync flag is per-table.** On-disk bookkeeping schema
+>   goes to **v3** (new `__localdb_snapshot_state`, upgraded in place on open); wire-compatible with
+>   `0.1.x`/`0.2.x` peers, and `lib_schema_version` deliberately stays `1` because it is compared
+>   fail-closed for equality.
+>   - **`RegisterReplicated(table, baselineExistingRows: true)` now flags a resync for THAT TABLE.**
+>     Through `0.2.x` the flag was one bit for the whole database, so baselining one table
+>     re-streamed every registered table in full — and on a pair enabling replication late, both
+>     nodes did that to each other at once. A snapshot now streams only the flagged tables.
+>   - **The database-wide flag survives, unchanged, for the cases that genuinely need it**: an oplog
+>     cap prune or an impossible peer ack loses deltas for the oplog as a whole, so nothing short of
+>     every table is sound. What moved is only where a *baseline* records its debt.
+>   - **Additive wire extension, negotiated by capability, not by version.** `Handshake` gains
+>     `supports_partial_snapshot`; `SnapshotBegin` gains a `tables` scope where **empty means "all
+>     tables"** — which is precisely what a pre-`0.3.0` sender emits and what a pre-`0.3.0` receiver
+>     assumes. A peer that does not advertise the capability is served a **full** snapshot, because a
+>     scoped one would make it advance `last_applied_remote_seq` past deltas it never received.
+>   - **A partial snapshot moves no watermark on either side.** The receiver keeps
+>     `last_applied_remote_seq` (it is one number for the whole peer oplog, and the untouched tables'
+>     deltas still sit below the snapshot's as-of); the sender keeps pumping from where it was. The
+>     overlap re-delivers rows the snapshot already carried, which LWW absorbs.
+>   - **Snapshot debt now clears on acknowledgement, not on enqueue** (new `SnapshotAck`, exchanged
+>     only between `0.3.0` peers). A crash part-way through a snapshot re-streams the same scope on
+>     reconnect instead of losing it. Against a pre-`0.3.0` peer the `0.2.x` clear-on-enqueue
+>     behaviour is kept — there is no ack to wait for.
+> - `0.2.1` (2026-08-14) — two adversarial-review findings against `0.2.0`. Wire-compatible with
+>   `0.1.x`/`0.2.0` peers; no proto change, no `lib_schema_version` bump, no on-disk schema change.
+>   - **`DeregisterReplicated` no longer leaves the HLC without a crash anchor.** Clock recovery on
+>     open is `MAX(hlc_high_water, oplog HLCs, ledger HLCs, last_seen_hlc)`, and `hlc_high_water` was
+>     written only by graceful disposal — while deregistration deletes the table's oplog AND ledger
+>     rows. A crash after a deregister boot could therefore recover the clock below stamps the peer
+>     already holds, and LWW discards those as older. The current clock is now flushed inside the
+>     deregistration transaction, before the prune (and the flush is MAX-guarded, so it can never
+>     regress the anchor).
+>   - **The sync inbox is bounded (64 messages) instead of unbounded**, so inbound memory is capped
+>     at roughly `64 x MaxBatchBytes` rather than "whatever the peer sends" — the bad case being a
+>     mutual full snapshot of a large dataset, buffered whole on both nodes at once. A full inbox now
+>     stops draining the gRPC stream, which closes the peer's flow-control window.
+>   - **A snapshot this node owes is streamed alongside a running receive loop**, not from inside the
+>     handshake. That is what makes bounding safe: previously neither side drained its inbox while
+>     sending, so a mutual snapshot converged only because the inbox was unbounded. Interleaving was
+>     already safe (snapshot rows and deltas both merge through LWW, the watermark write is
+>     MAX-guarded) — only the ordering changed.
+>   - **Documented the cost of late baselining**: the snapshot flag is per-database, so baselining
+>     one table re-streams every registered table, in both directions on a late-enabled pair. Trim
+>     high-volume tables (event logs) before turning replication on late.
+> - `0.2.0` (2026-08-14) — performance + lifecycle work from the ScadaBridge architectural review.
+>   All of it is wire-compatible with `0.1.x` peers (no proto change, no `lib_schema_version` bump);
+>   the on-disk bookkeeping schema goes to **v2** and is upgraded in place on open.
+>   - **`DeregisterReplicated(table)`** — drops a table's capture triggers and prunes its oplog and
+>     row-version rows. A node that does not replicate should not be paying for capture on every
+>     write, and until now there was no way to stop once a file had been registered. Idempotent, and
+>     usable on triggers left behind by an earlier build.
+>   - **`RegisterReplicated(table, baselineExistingRows: true)`** — closes the late-opt-in hole: rows
+>     that predate registration were invisible to the peer forever, because the snapshot streamer
+>     pages from the row-version ledger and CDC never wrote them there. Opting in seeds the ledger at
+>     an LWW-safe floor (HLC `0`, this node's id) and flags a snapshot resync so they actually
+>     travel. Default is still off — the documented CDC contract is unchanged unless you ask.
+>   - **O(1) backlog depth** — `MAX(seq) - MAX(last_acked_seq, MIN(seq) - 1)` replaces a
+>     `COUNT(*)` over the whole unacked backlog that ran once a minute AND on every metrics scrape
+>     (measured 58 VM steps vs 100,020 on a 50k backlog).
+>   - **Byte-budget batching** — `MaxBatchBytes` (default 2 MB) bounds delta and snapshot messages by
+>     summed serialized size, with `MaxBatchSize` demoted to a secondary row cap. Consumers no longer
+>     have to tune the row count against their widest row to stay under the gRPC message limit.
+>     Inbound delta batches and acks already queued are coalesced into one transaction / one ack.
+>   - **`LwwApplier` prepares once per statement** instead of three SQL parses per applied row.
+>   - **Dropped `__localdb_oplog_hlc`** — an index no query ever used, costing a b-tree insert inside
+>     every captured consumer write. Existing files lose it on the next open.
 > - `0.1.3` (2026-07-21) — fixes the other half of the rebuilt-peer story: a rebuilt peer's OWN
 >   writes were silently dropped. `last_applied_remote_seq` is a watermark in the *peer's* seq
 >   space, and a rebuilt peer numbers from 1 again — so the healthy node advertised the old peer's
@@ -107,7 +234,8 @@ node still pushes its own deltas back. Only one side needs `PeerAddress`.
 | `PeerAddress` | `""` | gRPC address to dial. Empty ⇒ passive (initiator idles). |
 | `ApiKey` | `null` | Bearer token the initiator presents on its channel; `null` ⇒ no key header. |
 | `FlushInterval` | `250 ms` | How often the change pump flushes pending oplog deltas. |
-| `MaxBatchSize` | `500` | Max oplog entries per outbound batch (also the snapshot page size). |
+| `MaxBatchBytes` | `2 MB` | **Primary** batch cap: summed serialized bytes per outbound delta/snapshot message. Sized under the 4 MB gRPC default; a single row above the budget is still sent, alone. |
+| `MaxBatchSize` | `500` | **Secondary** row-count cap per outbound batch (also the snapshot DB page size). |
 | `MaxOplogRows` | `1_000_000` | Backlog ceiling; exceeding it flags a snapshot resync and prunes oldest rows. |
 | `MaxOplogAge` | `7 days` | Age ceiling; an unpruned entry older than this flags a snapshot resync. |
 | `TombstoneRetention` | `7 days` | How long deleted-row tombstones are kept before pruning. |
@@ -133,8 +261,15 @@ All options are validated at startup (`ZB.MOM.WW.Configuration` validators).
  delta/snapshot batches), so receiving never blocks on the ability to ack.
 - **Resync** — when a peer's watermark falls below the pruned oplog horizon, it is sent a
  full snapshot of the registered tables (streamed, keyset-paged) and deltas resume above
-  the snapshot's as-of seq. **Each node decides from its OWN state** whether it owes the
-  peer a snapshot.
+  the snapshot's as-of seq. When instead a single table was *baselined*, only that table is
+  streamed and **no watermark moves on either side** — a partial snapshot speaks for nothing
+  outside its own tables, so the deltas it overlaps are still pumped and merged by LWW.
+  **Each node decides from its OWN state** whether it owes the peer a snapshot, and a peer
+  that does not advertise `supports_partial_snapshot` is sent a full one. The snapshot is
+  streamed *after* the handshake, alongside a running receive loop, so a node sending one
+  still applies what the peer sends it — a mutual snapshot exchanges under flow control
+  instead of being buffered whole in memory. Between `0.3.0` peers the sender's flags are
+  cleared by the receiver's `SnapshotAck`, so a crash mid-snapshot re-streams it.
 - **Maintenance** — a background service on **both** roles enforces the oplog row/age caps
  and prunes expired tombstones every `MaintenanceInterval`.

@@ -160,14 +295,38 @@ All options are validated at startup (`ZB.MOM.WW.Configuration` validators).
  `RegisterReplicated`, because `json_object` cannot capture it). A *typeless* column can
  still hold a BLOB value at runtime, which then surfaces as a write-time error, not a
  registration error.
+- **A replicated table may not be the parent of a row-mutating `ON DELETE` foreign key**
+  (`CASCADE` / `SET NULL` / `SET DEFAULT` from any other table, or from itself). Applying a
+  remote row uses `INSERT OR REPLACE`, whose implicit delete fires that action with change
+  capture suppressed — the child rows would change on the applying node only and never
+  replicate, diverging the pair undetectably. `RegisterReplicated` refuses such a table
+  (`0.3.3`); use `ON DELETE NO ACTION`/`RESTRICT` and delete children explicitly, or leave
+  the parent unreplicated. The check runs at registration and reads the whole database's
+  foreign keys, so a child table created *after* registration is caught at the next startup,
+  not at the moment of the `CREATE TABLE`.
 - **CDC semantics** — rows that existed **before** `RegisterReplicated` are not captured
-  and are not snapshotted; only changes after registration are replicated. Re-baselining
-  pre-existing rows requires re-writing them.
+  and are not snapshotted; only changes after registration are replicated. **Opt out with
+  `RegisterReplicated(table, baselineExistingRows: true)`**, which seeds the row-version
+  ledger for existing rows at the baseline floor (HLC `0`) and flags a snapshot resync
+  **for that table** so they reach the peer. The floor is below every genuine stamp, so a baselined row loses
+  every comparison against a real remote write and wins only where the peer has no version
+  of that key at all. Two nodes that baseline the same key both hold HLC `0` and resolve by
+  the node-id tie-break — convergent, but arbitrary as to which content survives.
+- **A snapshot still ships whole tables.** The flag is per-table as of `0.3.0`, so baselining
+  one table streams that table and no other — but *that* table travels in full, and two
+  cases still stream **everything**: an oplog cap prune / impossible peer ack (delta
+  continuity is lost for the oplog as a whole), and any peer older than `0.3.0` (which
+  cannot be told a snapshot is partial). So **purge or trim high-volume tables such as event
+  logs before baselining them**, and expect one slow first sync. Bounded either way (batch
+  budget out, bounded inbox in).
 - **REAL ±Infinity** is carried in SQLite's JSON dialect on the wire (fine for the
  SQLite↔SQLite peers this library targets).
 - `AddZbLocalDb` / `AddZbLocalDbReplication` are **first-registration-wins** (idempotent).
- A **mutual simultaneous snapshot** buffers the peer's inbound snapshot in memory while
-  this node is still sending its own (acceptable for the small datasets this lib targets).
+- **Inbound buffering is bounded, not zero.** The duplex inbox holds up to 64 messages, so
+  peak inbound memory is roughly `64 x MaxBatchBytes` (~128 MB at the 2 MB default, far
+  less in practice — messages are only that large when the sender has that much backlog).
+  Lower `MaxBatchBytes` to lower the ceiling. Beyond the bound the stream stops being
+  drained and the peer is slowed by flow control.

 ## Observability

@@ -196,8 +355,28 @@ provider is wired or the poll threw, so a DB failure never reads as "0 backlog")

 ## Ops notes

- When the oplog row/age caps are exceeded, maintenance prunes to the ceiling and flags
-  `needs_snapshot` (disk safety over delta continuity); the peer then snapshot-resyncs.
+- When the oplog row/age caps are exceeded, maintenance prunes to the ceiling and flags the
+  **database-wide** `needs_snapshot` (disk safety over delta continuity); the peer then gets a
+  full snapshot. Per-table debt lives separately in `__localdb_snapshot_state` and is what
+  baselining writes.
+- **Upgrading a pair to `0.3.0`** needs no coordination: the two nodes negotiate scoped snapshots
+  by capability, so a `0.3.0` node paired with a `0.2.x` one simply keeps sending full snapshots
+  until both are upgraded. Rolling the file forward is one-way — a `0.2.x` build reopening a v3
+  file leaves `__localdb_snapshot_state` in place and ignores it, so any per-table debt recorded
+  by `0.3.0` is invisible to it (roll forward again, or re-baseline).
+- **Turning a table off** — `DeregisterReplicated(table)` drops its capture triggers and prunes
+  its oplog + ledger rows. It must be applied **symmetrically**: the handshake compares
+  registered-table digests fail-closed, so a node that drops a table its peer still
+  replicates stops syncing with a schema-mismatch error. Deregister on both nodes, or on
+  neither. Re-enabling later should pass `baselineExistingRows: true`, or the rows that
+  accumulated while capture was off stay invisible to the peer — but trim high-volume
+  tables first (see Limitations: baselining flags a full, all-tables snapshot). The HLC's
+  durable anchor is flushed in the same transaction, so a crash after a deregister boot
+  cannot bring the clock back below stamps peers already hold.
+- **On-disk schema upgrades** are applied in place when the file is opened, in one
+  transaction, and every step is safe to re-run. `schema_version` in `__localdb_meta` is
+  local-file state only — it is NOT the wire `lib_schema_version`, so a pair mid-upgrade
+  keeps syncing.
 - The **`__localdb_dead_letter`** table holds poison rows (bad JSON, null-row non-tombstone
  entries) and over-drift rejects (when `FailClosedOnDrift`). Inspect it to diagnose
  rejects.
@@ -208,13 +387,17 @@ provider is wired or the poll threw, so a DB failure never reads as "0 backlog")

 ## Testing

-144 tests, all offline / no live dependencies — unit (HLC ordering + restart
-persistence, trigger SQL generation, LWW decision table, watermark/pruning, digest
-computation), integration over two real SQLite DBs on an in-memory (and real) gRPC pipe
-(bidirectional convergence, delete/update races, offline-accumulate → reconnect,
-prune → snapshot resync), plus seeded randomized property tests (random interleaved op
-sequences → both DBs identical after quiesce). Family conventions apply
-(`TreatWarningsAsErrors`).
+199 tests, all offline / no live dependencies — unit (HLC ordering + restart
+persistence, deregistration's HLC anchor surviving an ungraceful reopen, trigger SQL
+generation, LWW decision table, watermark/pruning, digest computation, schema-version
+upgrade, deregistration idempotency, baseline-floor LWW safety, O(1) depth vs a seeded
+backlog, byte-budget chunking, bounded-inbox backpressure under a slow reader), integration
+over two real SQLite DBs on an in-memory (and real) gRPC pipe (bidirectional convergence,
+delete/update races, offline-accumulate → reconnect, prune → snapshot resync, mutual
+snapshot over a capacity-1 transport, single-table baseline streaming only that table,
+mixed-version interop where a peer without `supports_partial_snapshot` gets a full one,
+snapshot debt surviving until the peer's `SnapshotAck`), plus seeded randomized property tests (random interleaved op sequences → both DBs identical after quiesce).
+Family conventions apply (`TreatWarningsAsErrors`).

 Build / test from this directory:

@@ -16,9 +16,19 @@ set -euo pipefail
 : "${GITEA_NUGET_SOURCE:?set GITEA_NUGET_SOURCE to your Gitea NuGet feed URL}"
 : "${GITEA_NUGET_KEY:?set GITEA_NUGET_KEY to your Gitea access token}"

-dotnet pack -c Release -o ./artifacts
+# Named solution, named output, and a cd to the repo root, all for one reason: `dotnet pack`
+# with neither a project argument nor a fixed working directory packs whatever solution
+# happens to sit in the CALLER's directory — and then pushes it to a live feed under this
+# script's name. Invoked as `bash /path/to/ZB.MOM.WW.LocalDb/build/push.sh` from another
+# repo's checkout, the relative form packed and published 25 unrelated packages (recovered
+# with `dotnet nuget delete`, but only because the token happened to carry delete rights).
+# A publish script must be unambiguous about WHAT it publishes no matter where it is run.
+REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+cd "$REPO_ROOT"

-dotnet nuget push "./artifacts/*.nupkg" \
+dotnet pack ZB.MOM.WW.LocalDb.slnx -c Release -o "$REPO_ROOT/artifacts"
+
+dotnet nuget push "$REPO_ROOT/artifacts/*.nupkg" \
  --source "$GITEA_NUGET_SOURCE" \
  --api-key "$GITEA_NUGET_KEY" \
  --skip-duplicate
@@ -16,6 +16,7 @@ message SyncMessage {
    SnapshotBegin snapshot_begin = 5; // sender switches to snapshot mode
    SnapshotBatch snapshot_batch = 6;
    SnapshotComplete snapshot_complete = 7;
+    SnapshotAck snapshot_ack = 8;     // 0.3.0+; only sent when the peer advertised support
  }
 }

@@ -24,6 +25,13 @@ message Handshake {
  uint32 lib_schema_version = 2;
  repeated TableDigest tables = 3;    // fail-closed on any mismatch
  int64 last_applied_remote_seq = 4;  // highest of YOUR oplog seqs I have applied
+  // 0.3.0+ capability, additive: I understand a snapshot scoped to a SUBSET of the registered
+  // tables (SnapshotBegin.tables) and I will answer a completed snapshot with a SnapshotAck.
+  // A pre-0.3.0 peer leaves this false — proto3 default — and a 0.3.0 sender then promotes any
+  // scoped snapshot to a full one and clears its flags on enqueue, exactly as 0.2.x did. This is
+  // why lib_schema_version stays 1: it is compared fail-closed for EQUALITY, so bumping it would
+  // stop 0.1.x/0.2.x pairs syncing outright.
+  bool supports_partial_snapshot = 5;
 }
 message TableDigest { string table_name = 1; string digest = 2; }
 message HandshakeAck { string node_id = 1; bool snapshot_required = 2; }
@@ -40,7 +48,12 @@ message OplogEntry {
 message DeltaBatch { repeated OplogEntry entries = 1; }
 message DeltaAck { int64 applied_thru_seq = 1; }

-message SnapshotBegin { int64 as_of_seq = 1; }
+// tables: the SCOPE of this snapshot (0.3.0+). EMPTY means every registered table — a full
+// snapshot, which is the only thing a pre-0.3.0 sender ever emits and the only thing a
+// pre-0.3.0 receiver can safely assume. A non-empty list means the snapshot is PARTIAL, so the
+// receiver must NOT advance last_applied_remote_seq to as_of_seq (deltas below it for the tables
+// NOT listed have not been delivered); it merges the rows through LWW and leaves the watermark.
+message SnapshotBegin { int64 as_of_seq = 1; repeated string tables = 2; }
 message SnapshotBatch { string table_name = 1; repeated SnapshotRow rows = 2; }
 message SnapshotRow {
  string pk_json = 1;
@@ -50,3 +63,8 @@ message SnapshotRow {
  bool is_tombstone = 5;
 }
 message SnapshotComplete { int64 as_of_seq = 1; }
+// Sent by the RECEIVER once a snapshot has been durably applied, so the sender clears the
+// per-table snapshot flags only after the peer actually holds the rows (a crash mid-snapshot then
+// re-streams the same scope on reconnect). Only exchanged when both peers advertised
+// supports_partial_snapshot; a pre-0.3.0 pair keeps the 0.2.x clear-on-enqueue behaviour.
+message SnapshotAck { int64 as_of_seq = 1; }
@@ -12,19 +12,58 @@ namespace ZB.MOM.WW.LocalDb.Replication.Internal;
 internal static class GrpcSyncDuplex
 {
    /// <summary>
-    /// Wires <paramref name="reader"/> into an unbounded inbox channel drained by a background loop,
-    /// and exposes <paramref name="writer"/> as the duplex send. Returns the duplex plus the reader
-    /// loop's task so the caller can observe it. On clean stream end the inbox channel is completed;
-    /// on a reader fault it is completed WITH the exception — the session's handshake and receive
-    /// phases both await the inbox, so terminating it is what unblocks them when the call dies (they
-    /// would otherwise hang on a dead stream).
+    /// Inbound messages buffered between the transport and the session's receive loop.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// <b>Why bounded at all.</b> An unbounded inbox makes this process's memory a function of what
+    /// the PEER decides to send. The worst case is not hypothetical: a full snapshot streams every
+    /// row of every registered table, and both sides of a pair that enabled replication late
+    /// snapshot each other simultaneously — so an unbounded inbox holds an entire peer dataset (an
+    /// event log at its storage cap, say) in memory on both nodes at once. Bounding it converts that
+    /// into backpressure: the pump stops draining the gRPC stream, the HTTP/2 flow-control window
+    /// closes, and the SENDER slows down.
+    /// </para>
+    /// <para>
+    /// <b>Why 64.</b> Memory here is bounded by roughly capacity x
+    /// <see cref="ReplicationOptions.MaxBatchBytes"/> (the sender's own per-message budget), so the
+    /// ceiling is a configured number rather than a peer-controlled one. 64 is four times the
+    /// sender's outbound data-lane depth, so a healthy stream never reaches the bound and pays
+    /// nothing; it engages only when this node is genuinely slower than the peer for a sustained
+    /// burst, which is exactly when backpressure is wanted. Lower <c>MaxBatchBytes</c> to lower the
+    /// ceiling.
+    /// </para>
+    /// <para>
+    /// <b>What makes bounding safe.</b> Backpressure on the inbox deadlocks any design where both
+    /// peers can stop draining their inbox at the same time — mutual snapshots were exactly that
+    /// case while the send ran inside the handshake. <see cref="SyncSession"/> therefore starts its
+    /// receive loop BEFORE streaming a snapshot it owes, so the inbox is drained continuously for
+    /// the whole life of the session; see the remarks on its snapshot phase.
+    /// </para>
+    /// </remarks>
+    internal const int InboxCapacity = 64;
+
+    /// <summary>
+    /// Wires <paramref name="reader"/> into a bounded inbox channel (see
+    /// <see cref="InboxCapacity"/>) drained by a background loop, and exposes
+    /// <paramref name="writer"/> as the duplex send. Returns the duplex plus the reader loop's task
+    /// so the caller can observe it. On clean stream end the inbox channel is completed; on a reader
+    /// fault it is completed WITH the exception — the session's handshake and receive phases both
+    /// await the inbox, so terminating it is what unblocks them when the call dies (they would
+    /// otherwise hang on a dead stream).
    /// </summary>
    public static (SyncDuplex Duplex, Task ReaderTask) Create(
        IAsyncStreamReader<SyncMessage> reader,
        IAsyncStreamWriter<SyncMessage> writer,
        CancellationToken ct)
    {
-        var inbox = Channel.CreateUnbounded<SyncMessage>(new UnboundedChannelOptions { SingleWriter = true });
+        var inbox = Channel.CreateBounded<SyncMessage>(new BoundedChannelOptions(InboxCapacity)
+        {
+            // Wait, never Drop: every inbound message is protocol state (a delta the peer will not
+            // resend in-session, a snapshot frame, an ack). Dropping one is silent divergence.
+            FullMode = BoundedChannelFullMode.Wait,
+            SingleWriter = true,
+        });
        var readerTask = PumpInboundAsync(reader, inbox.Writer, ct);
        var duplex = new SyncDuplex
        {
@@ -44,7 +83,10 @@ internal static class GrpcSyncDuplex
        try
        {
            while (await reader.MoveNext(ct))
-                inbox.TryWrite(reader.Current);
+                // WriteAsync, not TryWrite: on a FULL bounded channel TryWrite returns false and the
+                // message is silently lost. Awaiting here is the backpressure — it stops draining the
+                // gRPC stream, which closes the peer's flow-control window.
+                await inbox.WriteAsync(reader.Current, ct);
        }
        catch (Exception ex)
        {
@@ -12,14 +12,29 @@ internal sealed record ApplyResult(long AppliedThruSeq, int Applied, int Discard
 /// Applies a peer's oplog batch under last-writer-wins in one transaction: the applying guard is
 /// raised so capture triggers skip these writes, per-entry LWW resolves against the local
 /// row-version (hlc, then node_id ordinal tie-break), poison rows are dead-lettered while the
-/// batch continues, and the remote watermark advances in the SAME transaction as the data so a
+/// batch continues — each entry's row write and ledger write sharing a savepoint so a dead-lettered
+/// entry leaves neither — and the remote watermark advances in the SAME transaction as the data so a
 /// crash-replay of the batch is a no-op (LWW's strict-greater compare loses on equal versions).
 /// This never writes <c>__localdb_oplog</c>: the 2-node topology does not re-forward applied deltas.
 /// </summary>
+/// <remarks>
+/// Every statement is issued through a <see cref="PreparedStatements"/> set held for the life of the
+/// batch, so each distinct statement is prepared ONCE and then rebound per row. The previous shape
+/// went through the generic <c>ILocalDbTransaction</c> façade, which builds a fresh
+/// <see cref="SqliteCommand"/> per call — three SQL parses per applied row (version lookup, the
+/// table's delete/upsert, the row-version write), on the path that drains a reconnect backlog.
+/// </remarks>
 internal sealed class LwwApplier(SqliteLocalDb db, Func<DateTimeOffset>? utcNow = null)
 {
    private readonly Func<DateTimeOffset> _utcNow = utcNow ?? (() => DateTimeOffset.UtcNow);

+    /// <summary>
+    /// Distinct statements prepared by the most recent <see cref="ApplyBatchAsync"/>. Test-only
+    /// observation: it is the direct evidence that statement count tracks the number of distinct
+    /// STATEMENTS rather than the number of rows.
+    /// </summary>
+    internal int LastBatchStatementCount { get; private set; }
+
    public async Task<ApplyResult> ApplyBatchAsync(IReadOnlyList<OplogEntryRecord> entries, CancellationToken ct)
    {
        if (entries.Count == 0)
@@ -42,13 +57,14 @@ internal sealed class LwwApplier(SqliteLocalDb db, Func<DateTimeOffset>? utcNow
            if (entry.Hlc > maxHlc) maxHlc = entry.Hlc;
        }

-        var sqlByTable = new Dictionary<string, (string Delete, string Upsert)>(StringComparer.Ordinal);
        var applied = 0;
        var discarded = 0;
        var deadLettered = 0;

-        await using var tx = await db.BeginTransactionAsync(ct);
-        await tx.ExecuteAsync("UPDATE __localdb_applying SET applying = 1 WHERE id = 1", null, ct);
+        await using var tx = await db.BeginPreparedTransactionAsync(ct);
+        using var sql = new PreparedStatements(tx.Connection, tx.Sqlite);
+
+        await sql.SetApplyingAsync(1, ct);

        foreach (var entry in entries)
        {
@@ -56,19 +72,16 @@ internal sealed class LwwApplier(SqliteLocalDb db, Func<DateTimeOffset>? utcNow
            // KEY table that silently inserts a phantom auto-rowid row — so it is poison, not a win.
            if (!entry.IsTombstone && entry.RowJson is null)
            {
-                await DeadLetterAsync(tx, entry, "non-tombstone entry with null row_json", utcNow, ct);
+                await sql.DeadLetterAsync(entry, "non-tombstone entry with null row_json", utcNow, ct);
                deadLettered++;
                continue;
            }

-            var local = await tx.QueryAsync(
-                "SELECT hlc, node_id FROM __localdb_row_version WHERE table_name = @t AND pk_json = @pk",
-                static r => (Hlc: r.GetInt64(0), NodeId: r.GetString(1)),
-                new { t = entry.TableName, pk = entry.PkJson }, ct);
+            var local = await sql.ReadLocalVersionAsync(entry.TableName, entry.PkJson, ct);

-            var remoteWins = local.Count == 0
-                || entry.Hlc > local[0].Hlc
-                || (entry.Hlc == local[0].Hlc && string.CompareOrdinal(entry.NodeId, local[0].NodeId) > 0);
+            var remoteWins = local is null
+                || entry.Hlc > local.Value.Hlc
+                || (entry.Hlc == local.Value.Hlc && string.CompareOrdinal(entry.NodeId, local.Value.NodeId) > 0);

            if (!remoteWins)
            {
@@ -76,32 +89,28 @@ internal sealed class LwwApplier(SqliteLocalDb db, Func<DateTimeOffset>? utcNow
                continue;
            }

-            if (!sqlByTable.TryGetValue(entry.TableName, out var sql))
-            {
-                sql = BuildTableSql(db.ReplicatedTables[entry.TableName]);
-                sqlByTable[entry.TableName] = sql;
-            }
+            var table = db.ReplicatedTables[entry.TableName];

+            // An entry is TWO statements — the row change and its ledger row — and SQLite aborts
+            // only the failing statement, so a fault between them used to commit a half-applied
+            // entry: a tombstone whose table row is gone while __localdb_row_version still calls it
+            // live (which SnapshotStreamer can only warn about and skip, so the peer never learns
+            // the row died), or an upsert whose new row carries the OLD version and therefore loses
+            // the next LWW comparison it should win. A per-entry savepoint makes the pair
+            // all-or-nothing without widening the poison catch to the whole batch.
+            //
+            // Opened OUTSIDE the try on purpose: a SAVEPOINT that failed opened nothing, so the
+            // handler's ROLLBACK TO would name a savepoint that does not exist and fault in turn.
+            await sql.EnterEntryAsync(ct);
            try
            {
                if (entry.IsTombstone)
-                    await tx.ExecuteAsync(sql.Delete, new { pk = entry.PkJson }, ct);
+                    await sql.DeleteAsync(table, entry.PkJson, ct);
                else
-                    await tx.ExecuteAsync(sql.Upsert, new { row = entry.RowJson }, ct);
+                    await sql.UpsertAsync(table, entry.RowJson!, ct);

-                await tx.ExecuteAsync(
-                    "INSERT OR REPLACE INTO __localdb_row_version " +
-                    "(table_name, pk_json, hlc, node_id, is_tombstone, tombstone_utc) " +
-                    "VALUES (@t, @pk, @hlc, @node, @tomb, @tut)",
-                    new
-                    {
-                        t = entry.TableName,
-                        pk = entry.PkJson,
-                        hlc = entry.Hlc,
-                        node = entry.NodeId,
-                        tomb = entry.IsTombstone ? 1 : 0,
-                        tut = entry.IsTombstone ? utcNow : null,
-                    }, ct);
+                await sql.WriteRowVersionAsync(entry, utcNow, ct);
+                await sql.ReleaseEntryAsync(ct);
                applied++;
            }
            catch (SqliteException ex)
@@ -109,60 +118,234 @@ internal sealed class LwwApplier(SqliteLocalDb db, Func<DateTimeOffset>? utcNow
                // Deliberately narrow: only statement-level SQLite errors (e.g. json_extract on
                // malformed row_json) are treated as per-entry poison — SQLite aborts just the
                // statement, not the transaction, so the batch continues after dead-lettering.
-                // Any other exception propagates and rolls back the whole batch (fail-closed).
-                await DeadLetterAsync(tx, entry, ex.Message, utcNow, ct);
+                // Any other exception propagates with the savepoint still open, and disposing the
+                // outer transaction unwinds the whole batch (fail-closed).
+                try
+                {
+                    await sql.RollbackEntryAsync(ct);
+                    await sql.DeadLetterAsync(entry, ex.Message, utcNow, ct);
+                }
+                catch (Exception recoveryFailure) when (recoveryFailure is not OperationCanceledException)
+                {
+                    // Recovering from the poison entry itself failed — a wedged connection, an I/O
+                    // error, a dead-letter INSERT that faults. The outcome is unchanged and still
+                    // fail-closed (nothing commits; disposing the batch transaction unwinds it),
+                    // but reporting only the recovery failure would throw away the ORIGINAL cause,
+                    // which is the one that says WHY this entry is poison. Both are carried out:
+                    // the original as InnerException (AggregateException surfaces the first inner
+                    // there), the recovery failure alongside it with its own stack intact.
+                    throw new AggregateException(
+                        $"Failed to dead-letter a poison oplog entry for table '{entry.TableName}' " +
+                        $"(seq {entry.Seq}): {recoveryFailure.Message}. The entry's own fault is the " +
+                        "inner exception. The batch is not committed.",
+                        ex, recoveryFailure);
+                }
+
                deadLettered++;
            }
        }

-        await tx.ExecuteAsync(
-            "UPDATE __localdb_peer_state SET " +
-            "last_applied_remote_seq = MAX(last_applied_remote_seq, @maxSeq), " +
-            "last_seen_hlc = MAX(last_seen_hlc, @maxHlc), last_sync_utc = @utc WHERE id = 1",
-            new { maxSeq, maxHlc, utc = utcNow }, ct);
+        await sql.AdvancePeerStateAsync(maxSeq, maxHlc, utcNow, ct);
+        await sql.SetApplyingAsync(0, ct);
+
+        LastBatchStatementCount = sql.StatementCount;

-        await tx.ExecuteAsync("UPDATE __localdb_applying SET applying = 0 WHERE id = 1", null, ct);
        await tx.CommitAsync(ct);

        db.Clock.Observe(maxHlc);
        return new ApplyResult(maxSeq, applied, discarded, deadLettered);
    }

-    private static Task DeadLetterAsync(
-        ILocalDbTransaction tx, OplogEntryRecord entry, string error, string utcNow, CancellationToken ct) =>
-        tx.ExecuteAsync(
-            "INSERT INTO __localdb_dead_letter " +
-            "(received_utc, table_name, pk_json, row_json, hlc, node_id, error) " +
-            "VALUES (@utc, @t, @pk, @row, @hlc, @node, @err)",
-            new
-            {
-                utc = utcNow,
-                t = entry.TableName,
-                pk = entry.PkJson,
-                row = entry.RowJson,
-                hlc = entry.Hlc,
-                node = entry.NodeId,
-                err = error,
-            }, ct);
-
-    private static (string Delete, string Upsert) BuildTableSql(ReplicatedTable table)
+    /// <summary>
+    /// The statement set one apply-batch runs on: each distinct statement is built and prepared on
+    /// first use, then rebound and re-executed per row. Statements whose SQL depends on the table
+    /// (delete / upsert) are cached per table; the rest are shared across all of them.
+    /// </summary>
+    /// <remarks>
+    /// Reusing the same <see cref="SqliteCommand"/> with unchanged <c>CommandText</c> is what makes
+    /// this prepare-once: Microsoft.Data.Sqlite holds the compiled <c>sqlite3_stmt</c> on the command
+    /// and only recompiles when the text changes. <see cref="SqliteCommand.Prepare"/> is called
+    /// explicitly anyway, so the intent is stated rather than inferred from an implementation detail.
+    /// </remarks>
+    private sealed class PreparedStatements(SqliteConnection connection, SqliteTransaction transaction) : IDisposable
    {
-        var name = Ident(table.Name);
+        private readonly List<SqliteCommand> _owned = [];
+        private readonly Dictionary<string, SqliteCommand> _deletes = new(StringComparer.Ordinal);
+        private readonly Dictionary<string, SqliteCommand> _upserts = new(StringComparer.Ordinal);

-        var pkMatch = string.Join(" AND ", table.PkColumns.Select(p => $"{Ident(p)} = json_extract(@pk, {JsonPath(p)})"));
-        var delete = $"DELETE FROM {name} WHERE {pkMatch}";
+        /// <summary>Name of the per-entry savepoint. Fixed: only one is ever open at a time.</summary>
+        private const string EntrySavepoint = "zb_lww_entry";

-        var columns = string.Join(", ", table.Columns.Select(c => Ident(c.Name)));
-        var values = string.Join(", ", table.Columns.Select(c => $"json_extract(@row, {JsonPath(c.Name)})"));
-        var upsert = $"INSERT OR REPLACE INTO {name} ({columns}) VALUES ({values})";
+        private SqliteCommand? _applying;
+        private SqliteCommand? _enterEntry;
+        private SqliteCommand? _releaseEntry;
+        private SqliteCommand? _rollbackEntry;
+        private SqliteCommand? _lookup;
+        private SqliteCommand? _rowVersion;
+        private SqliteCommand? _deadLetter;
+        private SqliteCommand? _peerState;

-        return (delete, upsert);
+        /// <summary>How many distinct statements this set has prepared.</summary>
+        public int StatementCount => _owned.Count;
+
+        public Task SetApplyingAsync(int value, CancellationToken ct)
+        {
+            var cmd = _applying ??= Create(
+                "UPDATE __localdb_applying SET applying = @applying WHERE id = 1", "@applying");
+            cmd.Parameters["@applying"].Value = value;
+            return cmd.ExecuteNonQueryAsync(ct);
+        }
+
+        /// <summary>Opens the per-entry savepoint inside the batch transaction.</summary>
+        public Task EnterEntryAsync(CancellationToken ct)
+        {
+            var cmd = _enterEntry ??= Create($"SAVEPOINT {EntrySavepoint}");
+            return cmd.ExecuteNonQueryAsync(ct);
+        }
+
+        /// <summary>Commits the entry's work into the batch transaction (pops the savepoint).</summary>
+        public Task ReleaseEntryAsync(CancellationToken ct)
+        {
+            var cmd = _releaseEntry ??= Create($"RELEASE {EntrySavepoint}");
+            return cmd.ExecuteNonQueryAsync(ct);
+        }
+
+        /// <summary>
+        /// Undoes everything the entry did and pops the savepoint, leaving the batch transaction
+        /// open and usable. Two statements, because ROLLBACK TO rewinds to the savepoint but leaves
+        /// it ON the stack — without the RELEASE, every dead-lettered entry would nest another one.
+        /// </summary>
+        public async Task RollbackEntryAsync(CancellationToken ct)
+        {
+            var cmd = _rollbackEntry ??= Create($"ROLLBACK TO {EntrySavepoint}");
+            await cmd.ExecuteNonQueryAsync(ct);
+            await ReleaseEntryAsync(ct);
+        }
+
+        public async Task<(long Hlc, string NodeId)?> ReadLocalVersionAsync(
+            string tableName, string pkJson, CancellationToken ct)
+        {
+            var cmd = _lookup ??= Create(
+                "SELECT hlc, node_id FROM __localdb_row_version WHERE table_name = @t AND pk_json = @pk",
+                "@t", "@pk");
+            cmd.Parameters["@t"].Value = tableName;
+            cmd.Parameters["@pk"].Value = pkJson;
+
+            await using var reader = await cmd.ExecuteReaderAsync(ct);
+            if (!await reader.ReadAsync(ct))
+                return null;
+            return (reader.GetInt64(0), reader.GetString(1));
+        }
+
+        public Task DeleteAsync(ReplicatedTable table, string pkJson, CancellationToken ct)
+        {
+            if (!_deletes.TryGetValue(table.Name, out var cmd))
+                _deletes[table.Name] = cmd = Create(BuildDelete(table), "@pk");
+            cmd.Parameters["@pk"].Value = pkJson;
+            return cmd.ExecuteNonQueryAsync(ct);
+        }
+
+        public Task UpsertAsync(ReplicatedTable table, string rowJson, CancellationToken ct)
+        {
+            if (!_upserts.TryGetValue(table.Name, out var cmd))
+                _upserts[table.Name] = cmd = Create(BuildUpsert(table), "@row");
+            cmd.Parameters["@row"].Value = rowJson;
+            return cmd.ExecuteNonQueryAsync(ct);
+        }
+
+        public Task WriteRowVersionAsync(OplogEntryRecord entry, string utcNow, CancellationToken ct)
+        {
+            var cmd = _rowVersion ??= Create(
+                "INSERT OR REPLACE INTO __localdb_row_version " +
+                "(table_name, pk_json, hlc, node_id, is_tombstone, tombstone_utc) " +
+                "VALUES (@t, @pk, @hlc, @node, @tomb, @tut)",
+                "@t", "@pk", "@hlc", "@node", "@tomb", "@tut");
+            cmd.Parameters["@t"].Value = entry.TableName;
+            cmd.Parameters["@pk"].Value = entry.PkJson;
+            cmd.Parameters["@hlc"].Value = entry.Hlc;
+            cmd.Parameters["@node"].Value = entry.NodeId;
+            cmd.Parameters["@tomb"].Value = entry.IsTombstone ? 1 : 0;
+            cmd.Parameters["@tut"].Value = entry.IsTombstone ? utcNow : DBNull.Value;
+            return cmd.ExecuteNonQueryAsync(ct);
+        }
+
+        public Task DeadLetterAsync(OplogEntryRecord entry, string error, string utcNow, CancellationToken ct)
+        {
+            var cmd = _deadLetter ??= Create(
+                "INSERT INTO __localdb_dead_letter " +
+                "(received_utc, table_name, pk_json, row_json, hlc, node_id, error) " +
+                "VALUES (@utc, @t, @pk, @row, @hlc, @node, @err)",
+                "@utc", "@t", "@pk", "@row", "@hlc", "@node", "@err");
+            cmd.Parameters["@utc"].Value = utcNow;
+            cmd.Parameters["@t"].Value = entry.TableName;
+            cmd.Parameters["@pk"].Value = entry.PkJson;
+            cmd.Parameters["@row"].Value = (object?)entry.RowJson ?? DBNull.Value;
+            cmd.Parameters["@hlc"].Value = entry.Hlc;
+            cmd.Parameters["@node"].Value = entry.NodeId;
+            cmd.Parameters["@err"].Value = error;
+            return cmd.ExecuteNonQueryAsync(ct);
+        }
+
+        public Task AdvancePeerStateAsync(long maxSeq, long maxHlc, string utcNow, CancellationToken ct)
+        {
+            var cmd = _peerState ??= Create(
+                "UPDATE __localdb_peer_state SET " +
+                "last_applied_remote_seq = MAX(last_applied_remote_seq, @maxSeq), " +
+                "last_seen_hlc = MAX(last_seen_hlc, @maxHlc), last_sync_utc = @utc WHERE id = 1",
+                "@maxSeq", "@maxHlc", "@utc");
+            cmd.Parameters["@maxSeq"].Value = maxSeq;
+            cmd.Parameters["@maxHlc"].Value = maxHlc;
+            cmd.Parameters["@utc"].Value = utcNow;
+            return cmd.ExecuteNonQueryAsync(ct);
+        }
+
+        private SqliteCommand Create(string sql, params string[] parameterNames)
+        {
+            var cmd = connection.CreateCommand();
+            try
+            {
+                cmd.Transaction = transaction;
+                cmd.CommandText = sql;
+                foreach (var name in parameterNames)
+                    cmd.Parameters.Add(new SqliteParameter(name, DBNull.Value));
+                cmd.Prepare();
+                _owned.Add(cmd);
+                return cmd;
+            }
+            catch
+            {
+                cmd.Dispose();
+                throw;
+            }
+        }
+
+        public void Dispose()
+        {
+            foreach (var cmd in _owned)
+                cmd.Dispose();
+            _owned.Clear();
+        }
+
+        private static string BuildDelete(ReplicatedTable table)
+        {
+            var pkMatch = string.Join(
+                " AND ", table.PkColumns.Select(p => $"{Ident(p)} = json_extract(@pk, {JsonPath(p)})"));
+            return $"DELETE FROM {Ident(table.Name)} WHERE {pkMatch}";
+        }
+
+        private static string BuildUpsert(ReplicatedTable table)
+        {
+            var columns = string.Join(", ", table.Columns.Select(c => Ident(c.Name)));
+            var values = string.Join(", ", table.Columns.Select(c => $"json_extract(@row, {JsonPath(c.Name)})"));
+            return $"INSERT OR REPLACE INTO {Ident(table.Name)} ({columns}) VALUES ({values})";
+        }
+
+        // Local copy of TriggerSqlGenerator's convention (that helper lives in the core assembly).
+        private static string Ident(string name) => "\"" + name.Replace("\"", "\"\"") + "\"";
+
+        // A single-quoted JSON path with a quoted key member, so a column name with reserved
+        // characters still resolves: $."col". The key's double quotes are backslash-escaped per
+        // JSON path grammar.
+        private static string JsonPath(string column) => "'$.\"" + column.Replace("\"", "\\\"") + "\"'";
    }
-
-    // Local copy of TriggerSqlGenerator's convention (that helper lives in the core assembly).
-    private static string Ident(string name) => "\"" + name.Replace("\"", "\"\"") + "\"";
-
-    // A single-quoted JSON path with a quoted key member, so a column name with reserved characters
-    // still resolves: $."col". The key's double quotes are backslash-escaped per JSON path grammar.
-    private static string JsonPath(string column) => "'$.\"" + column.Replace("\"", "\\\"") + "\"'";
 }
@@ -24,8 +24,9 @@ internal sealed class OplogStore(ILocalDb db, ReplicationOptions options, Func<D

    /// <summary>
    /// Reads up to <paramref name="maxBatch"/> oplog entries with <c>seq &gt; afterSeq</c> in ascending
-    /// seq order. Caller contract: check <c>needs_snapshot</c> before trusting incremental reads —
-    /// after cap-pruning, reads silently resume past a gap.
+    /// seq order. Caller contract: this read resumes SILENTLY past a cap-pruned gap, so a caller
+    /// streaming incrementally must first confirm its position is still below the horizon — see
+    /// <see cref="GetMinSeqAsync"/>, which <c>SyncSession</c>'s pump checks on every iteration.
    /// </summary>
    public Task<IReadOnlyList<OplogEntryRecord>> ReadBatchAboveAsync(long afterSeq, int maxBatch, CancellationToken ct = default) =>
        db.QueryAsync(
@@ -70,12 +71,42 @@ internal sealed class OplogStore(ILocalDb db, ReplicationOptions options, Func<D
            "DELETE FROM __localdb_row_version WHERE is_tombstone = 1 AND tombstone_utc < @cutoff",
            new { cutoff = FormatUtc(_utcNow() - options.TombstoneRetention) }, ct);

-    /// <summary>Counts the unacked backlog: oplog entries above the peer's last-acked watermark.</summary>
+    /// <summary>
+    /// The unacked backlog — oplog entries above the peer's last-acked watermark — in O(1).
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// <b>Why not <c>COUNT(*)</c>.</b> The obvious query walks every unacked row, and this value is
+    /// read once a minute by maintenance AND on every metrics scrape. A node whose peer has been
+    /// down for a day therefore pays a full scan of a million-row backlog, repeatedly, precisely
+    /// when it is already under stress.
+    /// </para>
+    /// <para>
+    /// <b>The formula.</b> <c>MAX(seq) - MAX(last_acked_seq, MIN(seq) - 1)</c>, floored at 0. Both
+    /// MIN and MAX over <c>seq</c> are single b-tree seeks (it is the rowid, and SQLite's min/max
+    /// optimization turns each into one seek), and the peer-state read is a one-row lookup — so the
+    /// whole thing is three seeks regardless of backlog size.
+    /// </para>
+    /// <para>
+    /// <b>Why it is exact.</b> <c>seq</c> is AUTOINCREMENT, so retained rows are a contiguous run
+    /// wherever nothing has punched a hole in the middle — and no ordinary path punches holes: ack
+    /// pruning deletes <c>seq &lt;= last_acked_seq</c> (a prefix), the row cap deletes below
+    /// <c>MAX(seq) - keep + 1</c> (a prefix), and the age cap deletes below an HLC cutoff, which is
+    /// a prefix because HLCs are stamped in seq order. Clamping the floor to <c>MIN(seq) - 1</c> is
+    /// what keeps the answer exact after a prune has moved the horizon above the ack watermark;
+    /// without it a cap-pruned oplog would report its whole seq range forever and re-flag the cap on
+    /// every tick.
+    /// </para>
+    /// <para>
+    /// <b>The one inexact case</b> is <c>ILocalDb.DeregisterReplicated</c>, which deletes one
+    /// table's rows from wherever they sit and so can leave holes. The estimate then over-reports by
+    /// at most the number of holes, until the watermark advances past them. Over-reporting is the
+    /// safe direction: it can only trigger a snapshot resync earlier, never suppress one.
+    /// </para>
+    /// </remarks>
    public async Task<long> GetOplogDepthAsync(CancellationToken ct = default)
    {
-        var rows = await db.QueryAsync(
-            "SELECT COUNT(*) FROM __localdb_oplog WHERE seq > (SELECT last_acked_seq FROM __localdb_peer_state WHERE id = 1)",
-            static r => r.GetInt64(0), null, ct);
+        var rows = await db.QueryAsync(DepthSql, static r => r.GetInt64(0), null, ct);
        return rows[0];
    }

@@ -84,7 +115,7 @@ internal sealed class OplogStore(ILocalDb db, ReplicationOptions options, Func<D
    /// metric-observation callbacks (gauge scrape / status poll); mirrors
    /// <see cref="GetOplogDepthAsync"/>'s query. Deliberate tradeoff: synchronous I/O on the
    /// metrics-collection thread is acceptable here because WAL-mode readers never block (or are
-    /// blocked by) writers, the COUNT is indexed-cheap, and the worst case is bounded by
+    /// blocked by) writers, the query is three b-tree seeks, and the worst case is bounded by
    /// <c>BusyTimeoutMs</c> — all at scrape frequency, not on any data path. May throw if the db
    /// is disposing — callers on the observation path swallow and drop the sample.
    /// </summary>
@@ -92,11 +123,18 @@ internal sealed class OplogStore(ILocalDb db, ReplicationOptions options, Func<D
    {
        using var conn = db.CreateConnection();
        using var cmd = conn.CreateCommand();
-        cmd.CommandText =
-            "SELECT COUNT(*) FROM __localdb_oplog WHERE seq > (SELECT last_acked_seq FROM __localdb_peer_state WHERE id = 1)";
+        cmd.CommandText = DepthSql;
        return (long)cmd.ExecuteScalar()!;
    }

+    // Both depth accessors run exactly this; keeping it in one place is what makes "mirrors the
+    // async query" a fact rather than a comment that can rot.
+    internal const string DepthSql =
+        "SELECT MAX(0, " +
+        "COALESCE((SELECT MAX(seq) FROM __localdb_oplog), 0) - " +
+        "MAX((SELECT last_acked_seq FROM __localdb_peer_state WHERE id = 1), " +
+        "COALESCE((SELECT MIN(seq) FROM __localdb_oplog), 1) - 1))";
+
    /// <summary>
    /// The highest seq this node has ever produced: the oplog's max, or — once acked rows have been
    /// pruned away — the ack watermark that pruned them. Nothing we emit can exceed it, so a peer
@@ -111,6 +149,24 @@ internal sealed class OplogStore(ILocalDb db, ReplicationOptions options, Func<D
        return rows[0];
    }

+    /// <summary>
+    /// The oldest oplog seq still present — the delta horizon — or 0 when the oplog is empty.
+    /// </summary>
+    /// <remarks>
+    /// One b-tree seek (<c>seq</c> is the rowid and SQLite optimizes MIN into a seek), which is what
+    /// makes it affordable for the pump to check on every flush interval. Compared against a
+    /// session's sent watermark it answers the only question the pump has: is everything I have not
+    /// yet sent still here? Ack-pruning deletes <c>seq &lt;= last_acked_seq</c> and the peer can only
+    /// ack what it received, so in a healthy session the horizon never climbs past
+    /// <c>sentThruSeq + 1</c>; a higher one means a cap prune ate unsent deltas.
+    /// </remarks>
+    public async Task<long> GetMinSeqAsync(CancellationToken ct = default)
+    {
+        var rows = await db.QueryAsync(
+            "SELECT COALESCE(MIN(seq), 0) FROM __localdb_oplog", static r => r.GetInt64(0), null, ct);
+        return rows[0];
+    }
+
    /// <summary>The highest oplog seq present, or 0 when the oplog is empty.</summary>
    public async Task<long> GetMaxSeqAsync(CancellationToken ct = default)
    {
@@ -133,9 +189,9 @@ internal sealed class OplogStore(ILocalDb db, ReplicationOptions options, Func<D
        // Condition reads and prunes share the transaction so the decision cannot go stale (TOCTOU).
        await using var tx = await db.BeginTransactionAsync(ct);

-        var depth = (await tx.QueryAsync(
-            "SELECT COUNT(*) FROM __localdb_oplog WHERE seq > (SELECT last_acked_seq FROM __localdb_peer_state WHERE id = 1)",
-            static r => r.GetInt64(0), null, ct))[0];
+        // Same O(1) expression as GetOplogDepthAsync — the caps check is the per-minute reader that
+        // made the old COUNT(*) expensive in the first place.
+        var depth = (await tx.QueryAsync(DepthSql, static r => r.GetInt64(0), null, ct))[0];
        var overRows = depth > options.MaxOplogRows;

        var oldest = await tx.QueryAsync(
@@ -196,10 +252,61 @@ internal sealed class OplogStore(ILocalDb db, ReplicationOptions options, Func<D
        db.ExecuteAsync(
            "UPDATE __localdb_peer_state SET last_applied_remote_seq = 0, last_seen_hlc = 0 WHERE id = 1", null, ct);

-    /// <summary>Sets or clears the snapshot-resync-required flag.</summary>
+    /// <summary>
+    /// Sets or clears the DATABASE-WIDE snapshot-resync flag — the one that means "every registered
+    /// table". Set by the oplog-horizon paths (cap prune, impossible peer ack), which invalidate
+    /// delta continuity for all tables at once. Per-table flags live in
+    /// <c>__localdb_snapshot_state</c>; see <see cref="GetTablesNeedingSnapshotAsync"/>.
+    /// </summary>
    public Task SetNeedsSnapshotAsync(bool value, CancellationToken ct = default) =>
        db.ExecuteAsync("UPDATE __localdb_peer_state SET needs_snapshot = @v WHERE id = 1", new { v = value ? 1 : 0 }, ct);

+    /// <summary>
+    /// The tables individually flagged as owing the peer a snapshot (schema v3), in ordinal name
+    /// order. Independent of the database-wide flag: baselining one table flags only that table.
+    /// </summary>
+    public Task<IReadOnlyList<string>> GetTablesNeedingSnapshotAsync(CancellationToken ct = default) =>
+        db.QueryAsync(
+            "SELECT table_name FROM __localdb_snapshot_state WHERE needs_snapshot <> 0 ORDER BY table_name",
+            static r => r.GetString(0), null, ct);
+
+    /// <summary>
+    /// Clears the snapshot debt a completed snapshot discharged: the named tables when
+    /// <paramref name="tables"/> is non-empty, otherwise <b>everything</b> — the database-wide flag
+    /// and every per-table row, because a full snapshot subsumes each of them.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// Rows are deleted rather than zeroed: absence is the resting state, and keeping a row per
+    /// table that has ever been baselined would accumulate bookkeeping that means nothing. Runs in
+    /// one transaction so a full snapshot's two clears cannot half-apply and leave the pair
+    /// re-snapshotting on every reconnect.
+    /// </para>
+    /// <para>
+    /// A scoped clear deliberately does NOT touch the database-wide flag. If the cap enforcer
+    /// flagged a whole-database resync while this snapshot was in flight, that debt is real and
+    /// unrelated — the snapshot only carried some of the tables — so it must survive. The full
+    /// clear may take it down because a full snapshot carries every row state as of its as-of seq,
+    /// which is exactly what the pruned deltas would have delivered.
+    /// </para>
+    /// </remarks>
+    public async Task ClearSnapshotDebtAsync(IReadOnlyList<string> tables, CancellationToken ct = default)
+    {
+        await using var tx = await db.BeginTransactionAsync(ct);
+        if (tables.Count == 0)
+        {
+            await tx.ExecuteAsync("UPDATE __localdb_peer_state SET needs_snapshot = 0 WHERE id = 1", null, ct);
+            await tx.ExecuteAsync("DELETE FROM __localdb_snapshot_state", null, ct);
+        }
+        else
+        {
+            foreach (var table in tables)
+                await tx.ExecuteAsync(
+                    "DELETE FROM __localdb_snapshot_state WHERE table_name = @t", new { t = table }, ct);
+        }
+        await tx.CommitAsync(ct);
+    }
+
    /// <summary>
    /// Records the highest remote seq applied locally plus the peer HLC observed with it; also stamps
    /// <c>last_sync_utc</c>. Both watermarks are MAX-guarded (monotonic): a snapshot completing with a
@@ -19,21 +19,66 @@ internal interface ISnapshotApplier
 }

 /// <summary>
-/// Streams a full snapshot of every registered table to a peer that has fallen behind the pruned
-/// oplog horizon. Reads the authoritative per-row version from <c>__localdb_row_version</c> joined
+/// Which tables a snapshot covers. An EMPTY <see cref="Tables"/> is the wire's own encoding of
+/// "all registered tables" (see <c>SnapshotBegin.tables</c>) and is what every pre-0.3.0 peer both
+/// sends and expects, so it is deliberately the same value here rather than a separate flag.
+/// </summary>
+/// <remarks>
+/// The distinction is not cosmetic: a FULL snapshot carries every row state as of its
+/// <c>as_of_seq</c>, so the receiver may jump its watermark there and the sender may skip the
+/// deltas it covered. A PARTIAL one carries only the listed tables, so neither side may move a
+/// watermark that also speaks for the tables it left out.
+/// </remarks>
+internal sealed record SnapshotScope(IReadOnlyList<string> Tables)
+{
+    /// <summary>Every registered table — the only scope a pre-0.3.0 peer can be sent.</summary>
+    public static readonly SnapshotScope Full = new([]);
+
+    public bool IsPartial => Tables.Count > 0;
+
+    public bool Includes(string tableName) =>
+        Tables.Count == 0 || Tables.Contains(tableName, StringComparer.Ordinal);
+}
+
+/// <summary>
+/// Streams a snapshot of the tables in the requested <see cref="SnapshotScope"/> — every registered
+/// table for a peer that has fallen behind the pruned oplog horizon, or just the baselined ones for
+/// a scoped resync. Reads the authoritative per-row version from <c>__localdb_row_version</c> joined
 /// to each live table, pages it into <see cref="SnapshotBatch"/> messages through the gated data
 /// lane, and brackets them with SnapshotBegin/SnapshotComplete carrying the max oplog seq as-of.
 /// </summary>
+/// <remarks>
+/// Clearing the snapshot debt is deliberately NOT done here: whether the flags may be cleared at all
+/// depends on the peer's capability (a 0.3.0 peer acknowledges the snapshot, a pre-0.3.0 one cannot),
+/// which is session state. <see cref="SyncSession"/> owns that decision; this type only streams.
+/// </remarks>
 internal sealed class SnapshotStreamer(SqliteLocalDb db, OplogStore store, ReplicationOptions options, ILogger logger)
 {
-    /// <summary>Streams the full snapshot; returns the as-of oplog seq it covers, so the session can skip already-covered tail deltas.</summary>
-    public async Task<long> SendAsync(Func<SyncMessage, CancellationToken, Task> send, CancellationToken ct)
+    /// <summary>
+    /// Streams the snapshot for <paramref name="scope"/>; returns the as-of oplog seq it covers, so
+    /// the session can skip already-covered tail deltas (a FULL snapshot only — a partial one speaks
+    /// for nothing outside its tables).
+    /// </summary>
+    public async Task<long> SendAsync(SnapshotScope scope, Func<SyncMessage, CancellationToken, Task> send, CancellationToken ct)
    {
-        var asOfSeq = await store.GetMaxSeqAsync(ct);
-        await send(new SyncMessage { SnapshotBegin = new SnapshotBegin { AsOfSeq = asOfSeq } }, ct);
+        // The as-of is the highest seq this snapshot can stand in for. MAX(seq) alone is wrong once
+        // ack-pruning has emptied the oplog: it reads 0, and a receiver whose watermark genuinely IS
+        // 0 (a rebuilt node) is then never lifted by the snapshot — its MAX-guarded watermark stays
+        // 0, the empty-oplog horizon rule (streamable from last_acked_seq + 1) sees a gap on the
+        // next handshake, and the whole database is re-streamed on every reconnect until this node
+        // happens to write again. Every delta pruned below last_acked_seq is already folded into the
+        // rows this snapshot carries, so last_acked_seq is the floor the as-of must carry.
+        var asOfSeq = Math.Max(
+            await store.GetMaxSeqAsync(ct),
+            (await store.GetPeerStateAsync(ct)).LastAckedSeq);
+        var begin = new SnapshotBegin { AsOfSeq = asOfSeq };
+        begin.Tables.AddRange(scope.Tables);
+        await send(new SyncMessage { SnapshotBegin = begin }, ct);

        var totalRows = 0L;
-        foreach (var table in db.ReplicatedTables.Values.OrderBy(t => t.Name, StringComparer.Ordinal))
+        foreach (var table in db.ReplicatedTables.Values
+                     .Where(t => scope.Includes(t.Name))
+                     .OrderBy(t => t.Name, StringComparer.Ordinal))
        {
            var sql = BuildPageSql(table);
            // Keyset cursor (pk_json > last seen), not LIMIT/OFFSET: concurrent inserts/prunes
@@ -57,7 +102,13 @@ internal sealed class SnapshotStreamer(SqliteLocalDb db, OplogStore store, Repli
                    break;
                lastPk = rows[^1].PkJson;

+                // One DB page may span several messages: the page is bounded by ROW COUNT, but a
+                // message is bounded by BYTES (ReplicationOptions.MaxBatchBytes) — the same budget
+                // the delta pump honours, and for the same reason. Wide rows are exactly why a
+                // snapshot is the easiest way to exceed the transport's message limit: it ships
+                // whole rows, all of them, back to back.
                var batch = new SnapshotBatch { TableName = table.Name };
+                var bytes = 0L;
                foreach (var row in rows)
                {
                    if (!row.IsTombstone && row.RowJson is null)
@@ -79,11 +130,25 @@ internal sealed class SnapshotStreamer(SqliteLocalDb db, OplogStore store, Repli
                    };
                    if (row.RowJson is not null)
                        snapshotRow.RowJson = row.RowJson;
+
+                    var size = RowSize(snapshotRow);
+                    // Flush before adding, never after: a message keeps at least one row however
+                    // large, so an oversized row goes out alone rather than stalling the snapshot.
+                    if (batch.Rows.Count > 0 && bytes + size > options.MaxBatchBytes)
+                    {
+                        await send(new SyncMessage { SnapshotBatch = batch }, ct);
+                        totalRows += batch.Rows.Count;
+                        batch = new SnapshotBatch { TableName = table.Name };
+                        bytes = 0;
+                    }
                    batch.Rows.Add(snapshotRow);
+                    bytes += size;
                }
                if (batch.Rows.Count > 0)
+                {
                    await send(new SyncMessage { SnapshotBatch = batch }, ct);
-                totalRows += batch.Rows.Count;
+                    totalRows += batch.Rows.Count;
+                }

                if (rows.Count < options.MaxBatchSize)
                    break;
@@ -91,24 +156,29 @@ internal sealed class SnapshotStreamer(SqliteLocalDb db, OplogStore store, Repli
        }

        await send(new SyncMessage { SnapshotComplete = new SnapshotComplete { AsOfSeq = asOfSeq } }, ct);
-        // The sender owed this snapshot (its own needs_snapshot); clearing it here — after the whole
-        // snapshot is on the wire — stops it re-streaming on the next reconnect. Narrow window: the
-        // messages are enqueued on the data lane, not yet transport-delivered, so a crash/drop after
-        // this clear could lose the snapshot with the flag already down. The reconnect handshake's
-        // gap check (peer watermark below the pruned horizon) re-detects that in every case where the
-        // oplog is non-empty; the sole undetectable corner is an EMPTY oplog (nothing to gap-check),
-        // where the peer stays behind until new local writes recreate a horizon. Accepted: closing it
-        // would need transport-level delivery acks for snapshot frames.
-        await store.SetNeedsSnapshotAsync(false, ct);
-        logger.LogInformation("Snapshot sent (as_of_seq {AsOfSeq}, {Rows} rows).", asOfSeq, totalRows);
+        logger.LogInformation(
+            "Snapshot sent (as_of_seq {AsOfSeq}, {Rows} rows, scope {Scope}).",
+            asOfSeq, totalRows, scope.IsPartial ? string.Join(",", scope.Tables) : "all");
        return asOfSeq;
    }

+    /// <summary>
+    /// Serialized bytes one row contributes to its batch message: the row's own encoding plus the
+    /// repeated field's tag and length prefix, as protobuf itself computes them.
+    /// </summary>
+    private static long RowSize(SnapshotRow row) =>
+        1 + Google.Protobuf.CodedOutputStream.ComputeMessageSize(row);
+
    // Pages one table's rows by joining the row-version ledger to the live table on the PK columns
    // extracted from pk_json. Rows present in the live table but absent from __localdb_row_version
-    // (data written BEFORE the table was registered for replication) are intentionally NOT snapshotted:
-    // they were never captured, consistent with CDC semantics — a full re-registration is required to
-    // baseline them.
+    // (data written BEFORE the table was registered for replication) are NOT snapshotted: they were
+    // never captured, consistent with CDC semantics.
+    //
+    // That is a real hole for a node that starts replicating LATE — its whole pre-existing dataset is
+    // invisible to the peer, silently and permanently — and the supported way to close it is to
+    // register with `ILocalDb.RegisterReplicated(table, baselineExistingRows: true)`, which seeds the
+    // ledger at an LWW-safe floor and flags the resync that carries the rows here. Nothing changes on
+    // this side: baselined rows are ordinary ledger rows and page like any other.
    private static string BuildPageSql(ReplicatedTable table)
    {
        var name = Ident(table.Name);
@@ -149,6 +219,7 @@ internal sealed class SnapshotApplier : ISnapshotApplier
    private long _asOfSeq;
    private long _rows;
    private long _maxHlc;
+    private bool _partial;

    // db is part of the collaborator's canonical shape (mirrors the streamer) even though the merge
    // reaches the database through the applier and store; kept for construction symmetry in the factory.
@@ -165,7 +236,13 @@ internal sealed class SnapshotApplier : ISnapshotApplier
        _asOfSeq = begin.AsOfSeq;
        _rows = 0;
        _maxHlc = 0;
-        _logger.LogInformation("Snapshot receive begun (as_of_seq {AsOfSeq}).", _asOfSeq);
+        // A non-empty table list scopes the snapshot; empty (which is all a pre-0.3.0 sender can
+        // emit) means every registered table. Read here rather than on complete because
+        // SnapshotComplete does not repeat it — the receive loop guarantees a Begin precedes both.
+        _partial = begin.Tables.Count > 0;
+        _logger.LogInformation(
+            "Snapshot receive begun (as_of_seq {AsOfSeq}, scope {Scope}).",
+            _asOfSeq, _partial ? string.Join(",", begin.Tables) : "all");
        return Task.CompletedTask;
    }

@@ -187,9 +264,20 @@ internal sealed class SnapshotApplier : ISnapshotApplier

    public async Task OnCompleteAsync(SnapshotComplete complete, CancellationToken ct)
    {
+        // A PARTIAL snapshot may not move the inbound seq watermark. That watermark is one number
+        // for the whole peer oplog, and jumping it to as_of_seq would declare every table applied
+        // through that point — including the ones this snapshot deliberately left out, whose deltas
+        // below as_of_seq the sender would then never resend. Only the HLC is carried across (seq 0
+        // is a no-op against the MAX guard), so the clock still cannot regress below what the peer
+        // has stamped. The sender makes the mirror-image concession: it does not advance its own
+        // sent-thru watermark past a partial snapshot, so those deltas are still pumped normally
+        // and merely re-deliver rows the snapshot already carried, which LWW absorbs.
+        var seq = _partial ? 0 : complete.AsOfSeq;
        // SetLastAppliedRemoteSeqAsync is MAX-guarded, so a delta applied mid-snapshot with a seq
        // above the snapshot's as-of (or an HLC above what the snapshot carried) is never regressed.
-        await _store.SetLastAppliedRemoteSeqAsync(complete.AsOfSeq, _maxHlc, ct);
-        _logger.LogInformation("Snapshot receive complete (as_of_seq {AsOfSeq}, {Rows} rows).", complete.AsOfSeq, _rows);
+        await _store.SetLastAppliedRemoteSeqAsync(seq, _maxHlc, ct);
+        _logger.LogInformation(
+            "Snapshot receive complete (as_of_seq {AsOfSeq}, {Rows} rows, partial {Partial}).",
+            complete.AsOfSeq, _rows, _partial);
    }
 }
@@ -10,11 +10,13 @@ namespace ZB.MOM.WW.LocalDb.Replication.Internal;
 /// <summary>
 /// The symmetric 2-node sync protocol as one state machine: both the gRPC service and the gRPC
 /// client run this identical logic over a <see cref="SyncDuplex"/>. It performs the fail-closed
-/// handshake, then runs a bidirectional steady state — an outbound change pump and an inbound
-/// apply/ack loop — until the token is cancelled or the stream ends. All outbound traffic flows
-/// through a single writer task fed by two lanes: an unbounded CONTROL lane (handshake messages,
-/// acks) and a bounded DATA lane (delta/snapshot batches). Receiving therefore never blocks on the
-/// ability to write — acks always fit — while the bounded data lane is the pump's backpressure.
+/// handshake, then runs a bidirectional steady state until the token is cancelled or the stream
+/// ends: an outbound phase (the full snapshot this node owes, if any, then the change pump) and an
+/// inbound apply/ack loop, STARTED TOGETHER so the inbox keeps draining even while a snapshot is
+/// being sent. All outbound traffic flows through a single writer task fed by two lanes: an
+/// unbounded CONTROL lane (handshake messages, acks) and a bounded DATA lane (delta/snapshot
+/// batches). Receiving therefore never blocks on the ability to write — acks always fit — while the
+/// bounded data lane is the pump's backpressure.
 /// Local writes never block on the peer; the first fault tears down all loops and propagates.
 /// </summary>
 internal sealed class SyncSession
@@ -25,6 +27,10 @@ internal sealed class SyncSession
    // inbound batch rate (the peer's flow control), so its lane can safely be unbounded.
    private const int DataLaneCapacity = 16;

+    // How long a faulting session's teardown lets the writer keep flushing already-enqueued
+    // messages before the backstop cancel; only ever reached when the transport itself stalls.
+    private static readonly TimeSpan FaultFlushGrace = TimeSpan.FromSeconds(2);
+
    private readonly SqliteLocalDb _db;
    private readonly OplogStore _store;
    private readonly LwwApplier _applier;
@@ -45,6 +51,17 @@ internal sealed class SyncSession
    // single receive-loop task, so no synchronization is needed.
    private bool _snapshotReceiving;

+    // Whether the peer speaks the 0.3.0 snapshot extensions: scoped snapshots
+    // (SnapshotBegin.tables) and the SnapshotAck that answers a completed one. Written once during
+    // the handshake — which completes before either loop starts — then read-only.
+    private bool _peerSupportsPartialSnapshot;
+
+    // The scope of the snapshot this node is streaming, held until the peer's SnapshotAck discharges
+    // it; null on every session that owes none. Assigned on the pump task strictly BEFORE the
+    // snapshot's first message is enqueued and read on the receive loop when the ack lands, so the
+    // ordering is causal — volatile only to make the cross-thread publication explicit.
+    private SnapshotScope? _sentSnapshotScope;
+
    public SyncSession(
        SqliteLocalDb db, OplogStore store, LwwApplier applier, ReplicationOptions options, ILogger logger,
        Func<DateTimeOffset>? utcNow = null, LocalDbMetrics? metrics = null, SyncStatus? status = null)
@@ -61,11 +78,12 @@ internal sealed class SyncSession

    /// <summary>
    /// Invoked when THIS node owes the peer a snapshot (its own snapshot-required computation, not
-    /// the peer's) before deltas resume. Receives a gated send delegate that enqueues on the bounded
-    /// data lane (never the raw transport); returns the as-of oplog seq the snapshot covers so the
-    /// pump can skip already-covered tail deltas. Wired to <see cref="SnapshotStreamer.SendAsync"/>.
+    /// the peer's) before deltas resume. Receives the negotiated <see cref="SnapshotScope"/> and a
+    /// gated send delegate that enqueues on the bounded data lane (never the raw transport); returns
+    /// the as-of oplog seq the snapshot covers so the pump can skip already-covered tail deltas.
+    /// Wired to <see cref="SnapshotStreamer.SendAsync"/>.
    /// </summary>
-    public Func<Func<SyncMessage, CancellationToken, Task>, CancellationToken, Task<long>>? SnapshotSender { get; set; }
+    public Func<SnapshotScope, Func<SyncMessage, CancellationToken, Task>, CancellationToken, Task<long>>? SnapshotSender { get; set; }

    /// <summary>
    /// Consumes an inbound snapshot inline on the receive loop: SnapshotBegin opens the receive,
@@ -97,10 +115,19 @@ internal sealed class SyncSession
        var writer = WriterLoopAsync(duplex, control.Reader, data.Reader, linked.Token);
        try
        {
-            await HandshakeAsync(duplex, control.Writer, GatedDataSend, linked.Token);
+            var snapshotScope = await HandshakeAsync(duplex, control.Writer, linked.Token);

-            var pump = PumpLoopAsync(data.Writer, linked.Token);
+            // The receive loop starts BEFORE any snapshot this node owes is streamed. That ordering
+            // is load-bearing, not incidental: a snapshot sent from inside the handshake left the
+            // inbox undrained for its whole duration, so a MUTUAL snapshot (both sides flagged —
+            // e.g. both baselined an existing dataset) required an unbounded inbox to avoid
+            // deadlock, and therefore buffered a whole peer dataset in memory. With the receive loop
+            // running throughout, the peer's snapshot is applied as it arrives, the inbox can be
+            // bounded (see GrpcSyncDuplex.InboxCapacity), and mutual snapshots exchange under
+            // ordinary flow control. Interleaving is already safe by design: snapshot rows and
+            // deltas both merge through LWW, and the snapshot's watermark write is MAX-guarded.
            var receive = ReceiveLoopAsync(duplex, control.Writer, linked.Token);
+            var pump = SnapshotThenPumpLoopAsync(snapshotScope, GatedDataSend, data.Writer, linked.Token);
            var loops = new[] { pump, receive, writer };

            // The first COMPLETED task is the primary outcome. Cancel the siblings, observe every
@@ -117,15 +144,33 @@ internal sealed class SyncSession
        {
            // Handshake-phase faults (and the primary rethrow above) land here: the writer may
            // still be running, so stop and observe it before the primary exception surfaces.
-            linked.Cancel();
+            // It must be allowed to DRAIN, not be cancelled outright: the Handshake enqueued at
+            // the top of HandshakeAsync can still be sitting in the control lane (the enqueue →
+            // receive → validate → fault path has no yields, while the writer's wakeup waits in
+            // the thread-pool queue), and that message is the only thing that lets the peer
+            // compute its own fail-closed verdict — mismatch detection is symmetric, so this
+            // node's locally-thrown verdict reaches the peer only as the Handshake it transmits.
+            // Cancelling first killed it almost every time the peer's handshake was already
+            // buffered, leaving the healthy peer to misread a schema mismatch as a generic
+            // disconnect and reconnect-loop at the fast backoff. Completing the lanes lets the
+            // writer send what is already enqueued and exit on its own; the deferred cancel is
+            // the backstop that keeps a wedged transport from holding teardown hostage.
+            control.Writer.TryComplete();
+            data.Writer.TryComplete();
+            linked.CancelAfter(FaultFlushGrace);
            await DrainSecondaryAsync(writer);
            throw;
        }
    }

-    private async Task HandshakeAsync(
-        SyncDuplex duplex, ChannelWriter<SyncMessage> controlLane,
-        Func<SyncMessage, CancellationToken, Task> dataSend, CancellationToken ct)
+    /// <summary>
+    /// Runs the two-phase fail-closed handshake and returns the snapshot THIS node owes the peer —
+    /// null for none, otherwise the <see cref="SnapshotScope"/> to stream (its own computation from
+    /// its own state; the peer's answer drives the peer's send direction, symmetrically). Sending
+    /// that snapshot is deliberately NOT part of the handshake; see <see cref="RunAsync"/>.
+    /// </summary>
+    private async Task<SnapshotScope?> HandshakeAsync(
+        SyncDuplex duplex, ChannelWriter<SyncMessage> controlLane, CancellationToken ct)
    {
        var peerState = await _store.GetPeerStateAsync(ct);
        var localHandshake = BuildHandshake(peerState);
@@ -142,6 +187,11 @@ internal sealed class SyncSession

        ValidateHandshake(localHandshake, peerHandshake);

+        // Read before anything can act on it. A pre-0.3.0 peer leaves the field at its proto3
+        // default, so "false" covers both "explicitly unsupported" and "older build" without a
+        // version check — which is the point of advertising a capability rather than a version.
+        _peerSupportsPartialSnapshot = peerHandshake.SupportsPartialSnapshot;
+
        // A different node id on the far end means the peer was replaced, not merely restarted — a
        // rebuilt host, a restored-from-nothing volume. Its oplog numbers from 1 again, so the
        // watermark we hold (in the OLD peer's seq space) would silently swallow its first N
@@ -175,34 +225,78 @@ internal sealed class SyncSession
        _sentThruSeq = peerAppliedThruSeq;
        Interlocked.Exchange(ref _peerAckedSeq, peerAppliedThruSeq);

-        var snapshotRequired = await ComputeSnapshotRequiredAsync(peerHandshake, peerState, ct);
+        var snapshotScope = await ComputeSnapshotScopeAsync(peerHandshake, peerState, ct);
        await controlLane.WriteAsync(
-            new SyncMessage { HandshakeAck = new HandshakeAck { NodeId = _db.NodeId, SnapshotRequired = snapshotRequired } }, ct);
+            new SyncMessage { HandshakeAck = new HandshakeAck { NodeId = _db.NodeId, SnapshotRequired = snapshotScope is not null } }, ct);

        var peerAckMsg = await ReceiveOneAsync(duplex, ct);
        if (peerAckMsg.MsgCase != SyncMessage.MsgOneofCase.HandshakeAck)
            throw new InvalidOperationException(
                $"Replication protocol error: expected a HandshakeAck, got {peerAckMsg.MsgCase}.");

-        // We stream the snapshot when WE owe the peer one (computed from our pruned horizon /
-        // needs_snapshot), then it lands on the peer's receive loop. The peer's own flag drives ITS
-        // send direction symmetrically; both are safe to run because LWW merges either way.
-        // Memory profile of a MUTUAL snapshot: while we send here (receive loop not yet running),
-        // the peer's inbound frames — potentially its entire snapshot — buffer in the unbounded
-        // duplex inbox until our send completes. Worst case one full peer snapshot held in memory;
-        // acceptable for the small LocalDb datasets this library targets.
-        if (snapshotRequired)
+        return snapshotScope;
+    }
+
+    /// <summary>
+    /// The outbound steady state: stream the full snapshot this node owes the peer (if any), then
+    /// run the incremental delta pump. One task, because the ordering between them is a protocol
+    /// requirement — the pump must not put deltas on the data lane ahead of the snapshot, and it
+    /// starts from the as-of seq the snapshot covered.
+    /// </summary>
+    /// <remarks>
+    /// Split out of the handshake so the receive loop is already draining the inbox while this
+    /// streams (see <see cref="RunAsync"/>). Nothing else about the snapshot changes: it goes
+    /// through the same gated data lane, so it is still bounded by the writer's flow control rather
+    /// than racing the transport.
+    /// </remarks>
+    private async Task SnapshotThenPumpLoopAsync(
+        SnapshotScope? snapshotScope, Func<SyncMessage, CancellationToken, Task> dataSend,
+        ChannelWriter<SyncMessage> dataLane, CancellationToken ct)
+    {
+        if (snapshotScope is not null)
        {
            if (SnapshotSender is null)
                throw new NotSupportedException("snapshot required but no snapshot sender configured");
-            var snapshotAsOfSeq = await SnapshotSender(dataSend, ct);
+            // Published BEFORE the first frame is enqueued, so the peer's SnapshotAck can never
+            // arrive on the receive loop while this is still null.
+            Volatile.Write(ref _sentSnapshotScope, snapshotScope);
+            var snapshotAsOfSeq = await SnapshotSender(snapshotScope, dataSend, ct);
            _metrics?.RecordSnapshotSent();
            _status?.SetLastSyncUtc(_utcNow());
-            // The snapshot already carries every row state up to as-of; skipping the pump past it
-            // avoids re-sending tail deltas the peer would only discard via LWW.
-            if (snapshotAsOfSeq > _sentThruSeq)
+            // A FULL snapshot carries every row state up to as-of, so skipping the pump past it
+            // avoids re-sending tail deltas the peer would only discard via LWW. A PARTIAL one
+            // speaks only for its own tables, so the pump must still deliver every delta below
+            // as-of — the peer has not moved its watermark either (see SnapshotApplier.OnComplete).
+            if (!snapshotScope.IsPartial && snapshotAsOfSeq > _sentThruSeq)
                _sentThruSeq = snapshotAsOfSeq;
+
+            // A peer that cannot acknowledge the snapshot leaves only the 0.2.x option: clear once
+            // the frames are on the data lane. The window that opens (enqueued, not yet delivered —
+            // a crash there loses the snapshot with the debt already discharged) is why 0.3.0 peers
+            // wait for the ack instead. The reconnect handshake's gap check re-detects it whenever
+            // the oplog is non-empty; an EMPTY oplog remains the undetectable corner.
+            if (!_peerSupportsPartialSnapshot)
+                await ClearSentSnapshotDebtAsync(ct);
        }
+
+        await PumpLoopAsync(dataLane, ct);
+    }
+
+    /// <summary>
+    /// Discharges the snapshot debt this session's outbound snapshot covered, exactly once.
+    /// </summary>
+    /// <remarks>
+    /// Taking the scope out of the field before clearing makes this idempotent, which matters
+    /// because two paths reach it — the pre-0.3.0 clear-on-enqueue above and the peer's SnapshotAck
+    /// — and a peer that acknowledges anyway (a newer build than its handshake claimed) must not
+    /// clear a second time on top of flags a later baseline may already have re-raised.
+    /// </remarks>
+    private async Task ClearSentSnapshotDebtAsync(CancellationToken ct)
+    {
+        var scope = Interlocked.Exchange(ref _sentSnapshotScope, null);
+        if (scope is null)
+            return;
+        await _store.ClearSnapshotDebtAsync(scope.Tables, ct);
    }

    // The ONLY caller of duplex.Send. Control (handshake + acks) always preempts data so the peer's
@@ -246,20 +340,54 @@ internal sealed class SyncSession
        }
    }

+    /// <summary>
+    /// Reads oplog deltas above the sent watermark and ships them, splitting each read into as many
+    /// messages as <see cref="ReplicationOptions.MaxBatchBytes"/> requires.
+    /// </summary>
+    /// <remarks>
+    /// Row count alone cannot bound a message: rows are arbitrarily wide, so any row-count cap is
+    /// really a guess about the widest row, and guessing wrong produces a message the transport
+    /// rejects — replication then stops converging with a size error, not a data error. Budgeting by
+    /// summed serialized bytes bounds the thing that is actually limited, and lets the row cap go
+    /// back to meaning what it says.
+    /// </remarks>
    private async Task PumpLoopAsync(ChannelWriter<SyncMessage> dataLane, CancellationToken ct)
    {
        while (true)
        {
            ct.ThrowIfCancellationRequested();
+            // BEFORE the read, because the read itself cannot tell a pruned gap from an idle oplog.
+            ThrowIfPruneGap(await _store.GetMinSeqAsync(ct));
            var batch = await _store.ReadBatchAboveAsync(_sentThruSeq, _options.MaxBatchSize, ct);
            if (batch.Count > 0)
            {
-                var delta = new DeltaBatch();
-                foreach (var entry in batch)
-                    delta.Entries.Add(ToProto(entry));
-                // Blocking here on a full data lane IS the pump's backpressure.
-                await dataLane.WriteAsync(new SyncMessage { DeltaBatch = delta }, ct);
-                _sentThruSeq = batch[^1].Seq;
+                // The exact guard on the rows about to go out, for a prune that landed between the
+                // probe and the read. Free (the seq is already in hand) and strictly tighter.
+                ThrowIfPruneGap(batch[0].Seq);
+                var index = 0;
+                while (index < batch.Count)
+                {
+                    var delta = new DeltaBatch();
+                    var bytes = 0L;
+                    while (index < batch.Count)
+                    {
+                        var proto = ToProto(batch[index]);
+                        var size = EntrySize(proto);
+                        // The first entry of a message is taken however large it is: refusing an
+                        // oversized row would stall the stream on it permanently, whereas sending it
+                        // alone either succeeds or fails loudly at the transport with its real size.
+                        if (delta.Entries.Count > 0 && bytes + size > _options.MaxBatchBytes)
+                            break;
+                        delta.Entries.Add(proto);
+                        bytes += size;
+                        index++;
+                    }
+                    // Blocking here on a full data lane IS the pump's backpressure.
+                    await dataLane.WriteAsync(new SyncMessage { DeltaBatch = delta }, ct);
+                    // Advanced per MESSAGE, not per read: a fault mid-split must not re-send what
+                    // already went out, and must not skip what did not.
+                    _sentThruSeq = batch[index - 1].Seq;
+                }
                continue;
            }

@@ -267,23 +395,77 @@ internal sealed class SyncSession
        }
    }

+    /// <summary>
+    /// Faults the session when the oplog horizon has climbed above this session's read position —
+    /// i.e. a cap prune deleted deltas between <see cref="_sentThruSeq"/> and
+    /// <paramref name="oldestAvailableSeq"/>. An empty oplog (0) is not a gap.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// <b>Why a live session has to die for this.</b> <c>EnforceCapsAsync</c> sets
+    /// <c>needs_snapshot</c> in the same transaction as the prune, so a RECONNECT heals — the
+    /// handshake serves a full snapshot instead of deltas. Nothing, however, made the live session
+    /// notice: the pump reads <c>seq &gt; _sentThruSeq</c>, which silently resumes above the hole, and
+    /// the peer acks those later seqs as if the pair were converged. On a stable LAN with a slow but
+    /// alive peer that session can run for days, and the pair is durably divergent for all of it —
+    /// the flag that would fix it is only ever read at handshake time. Tearing down is what turns a
+    /// persisted "needs snapshot" into an actual snapshot.
+    /// </para>
+    /// <para>
+    /// <b>Why the horizon, not the flag.</b> <c>needs_snapshot</c> is also set for reasons that do
+    /// not invalidate THIS session's position (and a full snapshot this session just streamed clears
+    /// it only when the peer's <c>SnapshotAck</c> lands, which is after the pump has started).
+    /// Comparing the horizon against the sent watermark asks the precise question instead, so a
+    /// session is faulted when — and only when — rows it still owes the peer are gone.
+    /// </para>
+    /// </remarks>
+    private void ThrowIfPruneGap(long oldestAvailableSeq)
+    {
+        if (oldestAvailableSeq == 0 || oldestAvailableSeq <= _sentThruSeq + 1)
+            return;
+        throw new LocalDbOplogGapException(
+            $"Oplog deltas above the sent watermark {_sentThruSeq} were pruned (oldest surviving seq " +
+            $"{oldestAvailableSeq}): this session can no longer stream incrementally without skipping " +
+            "them. Ending the session so the reconnect handshake serves a snapshot resync.");
+    }
+
+    /// <summary>
+    /// Serialized bytes one entry contributes to its batch message: the entry's own encoding plus
+    /// the repeated field's tag and length prefix, as protobuf itself computes them.
+    /// </summary>
+    private static long EntrySize(OplogEntry entry) =>
+        1 + Google.Protobuf.CodedOutputStream.ComputeMessageSize(entry);
+
    private async Task ReceiveLoopAsync(
        SyncDuplex duplex, ChannelWriter<SyncMessage> controlLane, CancellationToken ct)
    {
-        while (await duplex.Inbox.WaitToReadAsync(ct))
+        // One-message lookahead. Coalescing means reading ahead until a message turns out NOT to be
+        // of the kind being coalesced, and a Channel cannot un-read — so that message is parked here
+        // and processed first on the next pass. Its presence also keeps the outer loop from waiting
+        // on the inbox when there is already work in hand.
+        SyncMessage? lookahead = null;
+
+        while (lookahead is not null || await duplex.Inbox.WaitToReadAsync(ct))
        {
-            while (duplex.Inbox.TryRead(out var msg))
+            while (true)
            {
+                var msg = lookahead;
+                lookahead = null;
+                if (msg is null)
+                {
+                    if (!duplex.Inbox.TryRead(out var read))
+                        break;
+                    msg = read;
+                }
+
                switch (msg.MsgCase)
                {
                    case SyncMessage.MsgOneofCase.DeltaBatch:
                        // Deltas apply normally even mid-snapshot: LWW makes the interleave safe.
-                        await HandleDeltaBatchAsync(controlLane, msg.DeltaBatch, ct);
+                        lookahead = await HandleCoalescedDeltasAsync(duplex, controlLane, msg.DeltaBatch, ct);
                        break;
                    case SyncMessage.MsgOneofCase.DeltaAck:
-                        var acked = msg.DeltaAck.AppliedThruSeq;
-                        InterlockedMax(ref _peerAckedSeq, acked);
-                        await _store.RecordPeerAckAsync(acked, ct);
+                        lookahead = await HandleCoalescedAcksAsync(duplex, msg.DeltaAck, ct);
                        break;
                    case SyncMessage.MsgOneofCase.SnapshotBegin:
                        if (SnapshotApplier is null)
@@ -305,6 +487,20 @@ internal sealed class SyncSession
                        _snapshotReceiving = false;
                        _metrics?.RecordSnapshotReceived();
                        _status?.SetLastSyncUtc(_utcNow());
+                        // Acknowledge only to a peer that asked for it. Sent AFTER the applier's
+                        // durable watermark write, so the ack means "I hold these rows", which is
+                        // what lets the sender drop its per-table flags. On the unbounded control
+                        // lane, like DeltaAck: an ack must never queue behind a data backlog.
+                        if (_peerSupportsPartialSnapshot)
+                            await controlLane.WriteAsync(
+                                new SyncMessage { SnapshotAck = new SnapshotAck { AsOfSeq = msg.SnapshotComplete.AsOfSeq } }, ct);
+                        break;
+                    case SyncMessage.MsgOneofCase.SnapshotAck:
+                        // The peer has durably applied the snapshot THIS node streamed, so the
+                        // tables it covered no longer owe one. Nothing to reconcile against the
+                        // ack's as-of seq: only one snapshot is ever sent per session, so the scope
+                        // held in _sentSnapshotScope is unambiguously the one being acknowledged.
+                        await ClearSentSnapshotDebtAsync(ct);
                        break;
                    default:
                        throw new InvalidOperationException(
@@ -314,19 +510,98 @@ internal sealed class SyncSession
        }
    }

-    private async Task HandleDeltaBatchAsync(ChannelWriter<SyncMessage> controlLane, DeltaBatch batch, CancellationToken ct)
+    /// <summary>
+    /// Applies the given delta batch together with any delta batches ALREADY waiting in the inbox,
+    /// as one transaction answered by one ack; returns the non-delta message that ended the run (to
+    /// be processed next), or null if the inbox merely ran dry.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// Byte-budgeting the sender produces more, smaller messages, and one apply transaction plus one
+    /// ack round-trip per message is the cost that would otherwise scale with that. Coalescing pays
+    /// it once per burst instead — and a burst is exactly what a reconnect after an outage delivers.
+    /// </para>
+    /// <para>
+    /// Only messages already queued are absorbed (<c>TryRead</c>, never a wait), so this adds no
+    /// latency: a trickle of one batch at a time coalesces nothing and behaves exactly as before.
+    /// The run stops at the sender's own caps, so the extra memory held is bounded by roughly one
+    /// sender batch beyond the budget — the overshoot of finishing whichever batch crosses it.
+    /// </para>
+    /// </remarks>
+    private async Task<SyncMessage?> HandleCoalescedDeltasAsync(
+        SyncDuplex duplex, ChannelWriter<SyncMessage> controlLane, DeltaBatch first, CancellationToken ct)
    {
-        if (batch.Entries.Count == 0)
+        var entries = new List<OplogEntry>(first.Entries);
+        var bytes = 0L;
+        foreach (var entry in first.Entries)
+            bytes += EntrySize(entry);
+
+        SyncMessage? lookahead = null;
+        while (entries.Count < _options.MaxBatchSize && bytes < _options.MaxBatchBytes
+               && duplex.Inbox.TryRead(out var next))
+        {
+            if (next.MsgCase != SyncMessage.MsgOneofCase.DeltaBatch)
+            {
+                lookahead = next;
+                break;
+            }
+            foreach (var entry in next.DeltaBatch.Entries)
+            {
+                entries.Add(entry);
+                bytes += EntrySize(entry);
+            }
+        }
+
+        await ApplyDeltaEntriesAsync(controlLane, entries, ct);
+        return lookahead;
+    }
+
+    /// <summary>
+    /// Records the given ack together with any acks ALREADY waiting in the inbox, as a single
+    /// watermark write; returns the non-ack message that ended the run, or null if the inbox ran dry.
+    /// </summary>
+    /// <remarks>
+    /// Exactly equivalent to recording each in turn, and strictly cheaper: <c>applied_thru_seq</c> is
+    /// monotonic and <c>RecordPeerAckAsync</c> is MAX-guarded, so every superseded ack would have
+    /// written a watermark that the final one overwrites and run a DELETE below its own horizon.
+    /// Unbounded coalescing is safe here — acks collapse into one number, accumulating nothing.
+    /// </remarks>
+    private async Task<SyncMessage?> HandleCoalescedAcksAsync(
+        SyncDuplex duplex, DeltaAck first, CancellationToken ct)
+    {
+        var acked = first.AppliedThruSeq;
+
+        SyncMessage? lookahead = null;
+        while (duplex.Inbox.TryRead(out var next))
+        {
+            if (next.MsgCase != SyncMessage.MsgOneofCase.DeltaAck)
+            {
+                lookahead = next;
+                break;
+            }
+            if (next.DeltaAck.AppliedThruSeq > acked)
+                acked = next.DeltaAck.AppliedThruSeq;
+        }
+
+        InterlockedMax(ref _peerAckedSeq, acked);
+        await _store.RecordPeerAckAsync(acked, ct);
+        return lookahead;
+    }
+
+    private async Task ApplyDeltaEntriesAsync(
+        ChannelWriter<SyncMessage> controlLane, IReadOnlyList<OplogEntry> batchEntries, CancellationToken ct)
+    {
+        if (batchEntries.Count == 0)
            return;

        var now = _utcNow();
        var nowMs = now.ToUnixTimeMilliseconds();
        var driftLimitMs = nowMs + (long)_options.MaxHlcDriftAhead.TotalMilliseconds;
-        var toApply = new List<OplogEntryRecord>(batch.Entries.Count);
+        var toApply = new List<OplogEntryRecord>(batchEntries.Count);
        var maxSeq = 0L;
        var driftDeadLettered = 0;

-        foreach (var proto in batch.Entries)
+        foreach (var proto in batchEntries)
        {
            if (proto.Seq > maxSeq) maxSeq = proto.Seq;
            var entry = FromProto(proto);
@@ -407,6 +682,9 @@ internal sealed class SyncSession
            NodeId = _db.NodeId,
            LibSchemaVersion = LibSchemaVersion,
            LastAppliedRemoteSeq = peerState.LastAppliedRemoteSeq,
+            // Unconditional: this build always understands a scoped snapshot and always answers a
+            // completed one. It is a statement about the code, not about configuration or state.
+            SupportsPartialSnapshot = true,
        };
        foreach (var table in _db.ReplicatedTables.Values.OrderBy(t => t.Name, StringComparer.Ordinal))
            handshake.Tables.Add(new TableDigest { TableName = table.Name, Digest = table.Digest });
@@ -437,11 +715,63 @@ internal sealed class SyncSession
            "Replication handshake rejected (fail-closed): " + string.Join("; ", parts) + ".");
    }

-    private async Task<bool> ComputeSnapshotRequiredAsync(Handshake peerHandshake, PeerState peerState, CancellationToken ct)
+    /// <summary>
+    /// Decides what snapshot this node owes the peer: null for none, <see cref="SnapshotScope.Full"/>
+    /// for every registered table, or a scoped snapshot naming just the tables that need one.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// The two sources of snapshot debt are genuinely different in reach, which is why they are
+    /// stored separately. An <b>oplog-horizon</b> gap — the peer's watermark below the oldest delta
+    /// still streamable, which is what a cap prune or an impossible peer ack leaves behind — means
+    /// deltas have been lost for the oplog as a whole, so nothing short of every table is sound. A
+    /// <b>baseline</b> means one table's pre-existing rows were never captured, and says nothing at
+    /// all about the others.
+    /// </para>
+    /// <para>
+    /// A scope that happens to name every registered table is returned as Full rather than as a
+    /// list: it covers exactly what a full snapshot covers, so it should also get a full snapshot's
+    /// watermark advance on both sides, and it needs no promotion for a pre-0.3.0 peer.
+    /// </para>
+    /// <para>
+    /// Flags naming tables that are not registered are ignored (they can only stream nothing).
+    /// <c>DeregisterReplicated</c> drops them with the table's ledger rows, so the surviving way to
+    /// hold one is to drop the SQL table without deregistering — which the handshake's fail-closed
+    /// digest comparison rejects for its own reasons.
+    /// </para>
+    /// </remarks>
+    private async Task<SnapshotScope?> ComputeSnapshotScopeAsync(Handshake peerHandshake, PeerState peerState, CancellationToken ct)
    {
-        if (peerState.NeedsSnapshot)
-            return true;
+        if (peerState.NeedsSnapshot || await HasHorizonGapAsync(peerHandshake, peerState, ct))
+            return SnapshotScope.Full;

+        var flagged = await _store.GetTablesNeedingSnapshotAsync(ct);
+        if (flagged.Count == 0)
+            return null;
+
+        var scoped = flagged.Where(_db.ReplicatedTables.ContainsKey).ToList();
+        if (scoped.Count == 0)
+            return null;
+        if (scoped.Count >= _db.ReplicatedTables.Count)
+            return SnapshotScope.Full;
+
+        // A peer that would read SnapshotBegin.tables as "all tables" and then advance its watermark
+        // past deltas this snapshot did not carry gets the full snapshot instead. Wider than asked
+        // for, but it is the only shape the older protocol can express correctly — and it is what
+        // 0.2.x did for every baseline.
+        if (!_peerSupportsPartialSnapshot)
+        {
+            _logger.LogInformation(
+                "Peer {PeerNodeId} does not support scoped snapshots; promoting the snapshot of {Tables} to a full one.",
+                PeerNodeId, string.Join(",", scoped));
+            return SnapshotScope.Full;
+        }
+
+        return new SnapshotScope(scoped);
+    }
+
+    private async Task<bool> HasHorizonGapAsync(Handshake peerHandshake, PeerState peerState, CancellationToken ct)
+    {
        var minRows = await _db.QueryAsync(
            "SELECT COALESCE(MIN(seq), 0) FROM __localdb_oplog", static r => r.GetInt64(0), null, ct);
        var minSeq = minRows[0];
@@ -6,7 +6,7 @@ namespace ZB.MOM.WW.LocalDb.Replication.Internal;
 /// <summary>
 /// Shared construction of a <see cref="SyncSession"/> and its dependencies (OplogStore + LwwApplier)
 /// from a <see cref="SqliteLocalDb"/> and the bound options. Both the service and client adapters use
-/// this so they build an identical session; snapshot hooks are left null (Task 12 fills them in).
+/// this so they build an identical session.
 /// </summary>
 internal static class SyncSessionFactory
 {
@@ -0,0 +1,23 @@
+namespace ZB.MOM.WW.LocalDb.Replication;
+
+/// <summary>
+/// Thrown by a live sync session when the oplog horizon has moved ABOVE the position that session
+/// is streaming from: a cap prune deleted deltas it had not yet sent, so every later delta it could
+/// ship would land on the far side of a hole the peer will never receive.
+/// </summary>
+/// <remarks>
+/// This is not a transport failure and not a data error — it is the session discovering that its own
+/// incremental position has become unsound. Nothing can repair it in-session (the deleted rows are
+/// gone), so the session tears itself down; the same prune set <c>needs_snapshot</c>, which makes the
+/// reconnect handshake serve a snapshot instead of deltas. Catching this and continuing would
+/// reintroduce exactly the silent divergence it exists to stop.
+/// </remarks>
+public sealed class LocalDbOplogGapException : Exception
+{
+    /// <summary>Creates the exception with a message describing the gap.</summary>
+    public LocalDbOplogGapException(string message) : base(message) { }
+
+    /// <summary>Creates the exception with a message and the underlying cause.</summary>
+    public LocalDbOplogGapException(string message, Exception innerException)
+        : base(message, innerException) { }
+}
@@ -21,8 +21,26 @@ public sealed class LocalDbSyncService : LocalDbSync.LocalDbSyncBase
    private readonly LocalDbMetrics? _metrics;
    private readonly SyncStatus? _status;

-    // 0 = idle, 1 = a stream is active. Interlocked-guarded across concurrent calls (singleton service).
-    private int _active;
+    // The stream currently holding the guard, or null when idle. Guarded by _gate (singleton service).
+    private readonly object _gate = new();
+    private Registration? _active;
+
+    /// <summary>
+    /// How long a displacing call waits for the incumbent it evicted to actually finish before
+    /// giving up and refusing itself. Test seam; the default is generous because eviction only has
+    /// to outrun the operator, not the network.
+    /// </summary>
+    internal TimeSpan TakeoverTimeout { get; set; } = TimeSpan.FromSeconds(10);
+
+    /// <summary>One in-flight <c>Sync</c> handler: its cancellation handle plus a completion signal
+    /// a displacing handler can await, so two sessions never run against the database at once.</summary>
+    private sealed class Registration(CancellationTokenSource cts)
+    {
+        public CancellationTokenSource Cts { get; } = cts;
+
+        public TaskCompletionSource Finished { get; } =
+            new(TaskCreationOptions.RunContinuationsAsynchronously);
+    }

    /// <summary>
    /// Creates the passive sync service. <paramref name="status"/> is taken as the public
@@ -41,21 +59,47 @@ public sealed class LocalDbSyncService : LocalDbSync.LocalDbSyncBase
        _status = status as SyncStatus;
    }

+    /// <summary>
+    /// Accepts one inbound sync stream, evicting any stream already holding the guard.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// <b>Newest wins.</b> The guard exists to SERIALIZE sessions — a node syncs with exactly one
+    /// peer, and two sessions against one database would fight over the same watermarks — not to
+    /// fence out an impostor; authentication is the interceptor's job, and an authenticated caller
+    /// on this key IS the peer by definition. Refusing the newcomer therefore protected nothing and
+    /// cost everything in the case that actually happens: the dialling node crashes or is
+    /// partitioned, this handler stays parked forever inside <c>MoveNext</c> on a half-open stream
+    /// that will never produce another byte, and every reconnect from the rebooted peer is rejected
+    /// <c>AlreadyExists</c> — potentially for as long as the pair stays idle, while this side keeps
+    /// reporting itself connected. The newest dialler is the one with a live link, so it takes over.
+    /// </para>
+    /// <para>
+    /// Eviction is cancel-then-await, not cancel-and-go: the incumbent's cancellation unwinds its
+    /// session and reader pump, and the newcomer waits for that to complete before starting its own,
+    /// so the "one session at a time" property the guard was built for still holds. An incumbent
+    /// that cannot be evicted within <see cref="TakeoverTimeout"/> is the one case where the old
+    /// answer is still right — the newcomer refuses itself rather than double-running.
+    /// </para>
+    /// </remarks>
    public override async Task Sync(
        IAsyncStreamReader<SyncMessage> requestStream,
        IServerStreamWriter<SyncMessage> responseStream,
        ServerCallContext context)
    {
-        if (Interlocked.CompareExchange(ref _active, 1, 0) != 0)
-            throw new RpcException(new Status(StatusCode.AlreadyExists, "a sync stream is already active"));
+        // The linked source exists so the reader pump is deterministically torn down on EVERY exit
+        // path — including a session fault for a local (non-transport) reason, where the call token
+        // never fires and the pump would otherwise keep reading a stream whose handler has exited.
+        // It is also the handle a later call cancels to displace this one.
+        using var linkedCts = CancellationTokenSource.CreateLinkedTokenSource(context.CancellationToken);
+        var mine = new Registration(linkedCts);

        try
        {
-            // The linked source exists so the reader pump is deterministically torn down on EVERY
-            // exit path — including a session fault for a local (non-transport) reason, where the
-            // call token never fires and the pump would otherwise keep reading a stream whose
-            // handler has exited.
-            using var linkedCts = CancellationTokenSource.CreateLinkedTokenSource(context.CancellationToken);
+            // Inside the try so the finally below always releases the guard — including when the
+            // takeover itself is abandoned because THIS caller gave up mid-handover.
+            await TakeOverAsync(mine, context.CancellationToken);
+
            var (duplex, readerTask) = GrpcSyncDuplex.Create(requestStream, responseStream, linkedCts.Token);
            var session = SyncSessionFactory.Create(_db, _options, _loggerFactory, _metrics, _status);
            _status?.SessionStarted();
@@ -81,13 +125,66 @@ public sealed class LocalDbSyncService : LocalDbSync.LocalDbSyncBase
        {
            // Client disconnect / server shutdown: the normal end of a long-lived streaming RPC.
        }
+        catch (OperationCanceledException)
+        {
+            // Displaced by a newer stream: an orderly handover, not a failure of this node.
+            _logger.LogInformation("Sync stream ended: displaced by a newer stream from the peer.");
+        }
        catch (LocalDbSchemaMismatchException ex)
        {
            throw new RpcException(new Status(StatusCode.FailedPrecondition, ex.Message));
        }
        finally
        {
-            Interlocked.Exchange(ref _active, 0);
+            lock (_gate)
+            {
+                if (ReferenceEquals(_active, mine))
+                    _active = null;
+            }
+            mine.Finished.TrySetResult();
+        }
+    }
+
+    /// <summary>
+    /// Claims the single-stream guard for <paramref name="mine"/>, cancelling and awaiting whatever
+    /// held it. Throws <c>AlreadyExists</c> only when the incumbent refuses to end.
+    /// </summary>
+    private async Task TakeOverAsync(Registration mine, CancellationToken ct)
+    {
+        Registration? displaced;
+        lock (_gate)
+        {
+            displaced = _active;
+            _active = mine;
+        }
+        if (displaced is null)
+            return;
+
+        _logger.LogWarning(
+            "A new sync stream arrived while one was already active; displacing the incumbent (newest wins).");
+        try
+        {
+            // The incumbent may already be disposing its own source between its finally and here.
+            displaced.Cts.Cancel();
+        }
+        catch (ObjectDisposedException) { }
+
+        try
+        {
+            await displaced.Finished.Task.WaitAsync(TakeoverTimeout, ct);
+        }
+        catch (TimeoutException)
+        {
+            // Could not evict it: two live sessions over one database is worse than a refused dial,
+            // so fall back to the historical answer — and hand the guard back to the incumbent.
+            lock (_gate)
+            {
+                if (ReferenceEquals(_active, mine))
+                    _active = displaced;
+            }
+            throw new RpcException(new Status(
+                StatusCode.AlreadyExists,
+                "a sync stream is already active and did not yield within the takeover timeout"));
        }
    }
 }
@@ -54,7 +54,7 @@ public sealed class MaintenanceBackgroundService : BackgroundService
                {
                    _snapshotFlagWarned = true;
                    _logger.LogWarning(
-                        "Oplog backlog/age caps exceeded: pruned to the ceiling and flagged needs_snapshot — the peer must snapshot-resync. Not re-logged until a clean tick.");
+                        "Oplog backlog/age caps exceeded: pruned to the ceiling and flagged needs_snapshot — the peer must snapshot-resync. A session streaming from below the new horizon ends itself on its next pump tick so the reconnect serves that snapshot. Not re-logged until a clean tick.");
                }
                else if (!snapshotFlagged)
                {
@@ -14,9 +14,44 @@ public sealed class ReplicationOptions
    /// <summary>How often the change pump flushes pending oplog deltas to the peer.</summary>
    public TimeSpan FlushInterval { get; set; } = TimeSpan.FromMilliseconds(250);

-    /// <summary>Maximum oplog entries sent per outbound batch.</summary>
+    /// <summary>
+    /// Secondary, row-count cap on an outbound batch (and the snapshot streamer's DB page size).
+    /// <see cref="MaxBatchBytes"/> is the primary cap.
+    /// </summary>
+    /// <remarks>
+    /// Historically this was the ONLY cap, which made it a proxy for a byte budget that consumers
+    /// had to tune by hand against their widest row — ScadaBridge pinned it to 16 because a 70 KB
+    /// <c>config_json</c> times the 500 default is ~35 MB against a 4 MB gRPC limit. With
+    /// <see cref="MaxBatchBytes"/> doing that job directly, this is back to being what its name
+    /// says: a bound on how many rows one message may carry.
+    /// </remarks>
    public int MaxBatchSize { get; set; } = 500;

+    /// <summary>
+    /// Primary cap on an outbound batch: the summed serialized size of the entries in one
+    /// <c>DeltaBatch</c> / <c>SnapshotBatch</c> message, in bytes. Default 2 MB.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// Sized at half the default 4 MB gRPC receive limit, so the message framing, the oneof wrapper
+    /// and any per-entry overhead the estimate does not model all fit inside the margin.
+    /// </para>
+    /// <para>
+    /// <b>A single row larger than the budget is still sent, alone.</b> Skipping it would stall the
+    /// stream permanently on one oversized row; sending it gives the transport the chance to accept
+    /// it (gRPC limits are configurable) and, if it refuses, a loud transport error naming the size
+    /// rather than silent non-convergence.
+    /// </para>
+    /// <para>
+    /// <b>Wire compatibility:</b> none required. Batch size is a sender-side choice the protocol
+    /// already leaves free — <c>DeltaBatch</c>/<c>SnapshotBatch</c> carry a repeated field with no
+    /// declared count — so a node on this version interoperates with a peer on any earlier one in
+    /// both directions, with no proto change and no <c>lib_schema_version</c> bump. (ScadaBridge
+    /// site pairs stop and start together regardless; this simply does not depend on it.)
+    /// </para>
+    /// </remarks>
+    public int MaxBatchBytes { get; set; } = 2 * 1024 * 1024;
+
    /// <summary>Backlog ceiling: exceeding it flags a snapshot resync and prunes the oldest oplog rows.</summary>
    public long MaxOplogRows { get; set; } = 1_000_000;

@@ -29,6 +64,34 @@ public sealed class ReplicationOptions
    /// <summary>Upper bound on the exponential reconnect backoff.</summary>
    public TimeSpan ReconnectBackoffMax { get; set; } = TimeSpan.FromSeconds(60);

+    /// <summary>
+    /// How long the initiator's HTTP/2 connection may sit without a keepalive PING before one is
+    /// sent. Default 60s.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// A sync stream is long-lived and, on a converged pair, almost entirely silent — which is
+    /// exactly the shape a half-open TCP connection hides behind. Without pings, a hard crash or a
+    /// partition of one node leaves the other holding a stream that will never produce another byte
+    /// and never fail either: the initiator sits in a session it believes is healthy, and the
+    /// passive node's handler sits inside its single-stream guard, refusing the reconnect that would
+    /// heal the pair. Pings turn that into a detected failure on both ends.
+    /// </para>
+    /// <para>
+    /// Sent with <c>KeepAlivePingPolicy.Always</c>, so they flow on an idle connection too — an
+    /// active-streams-only policy would not fire in precisely the silent-pair case above. Safe over
+    /// h2c: a PING is an HTTP/2 frame, nothing to do with TLS.
+    /// </para>
+    /// </remarks>
+    public TimeSpan KeepAlivePingDelay { get; set; } = TimeSpan.FromSeconds(60);
+
+    /// <summary>
+    /// How long the initiator waits for a keepalive PING ack before tearing the connection down.
+    /// Default 20s. Detection therefore lands within roughly
+    /// <see cref="KeepAlivePingDelay"/> + this.
+    /// </summary>
+    public TimeSpan KeepAlivePingTimeout { get; set; } = TimeSpan.FromSeconds(20);
+
    /// <summary>How often the maintenance service enforces the oplog caps and prunes expired tombstones.</summary>
    public TimeSpan MaintenanceInterval { get; set; } = TimeSpan.FromSeconds(60);

@@ -28,6 +28,8 @@ public sealed class ReplicationOptionsValidator : IValidateOptions<ReplicationOp
            failures.Add($"LocalDb:Replication:FlushInterval must be greater than zero; got {options.FlushInterval}.");
        if (options.MaxBatchSize <= 0)
            failures.Add($"LocalDb:Replication:MaxBatchSize must be greater than 0; got {options.MaxBatchSize}.");
+        if (options.MaxBatchBytes <= 0)
+            failures.Add($"LocalDb:Replication:MaxBatchBytes must be greater than 0; got {options.MaxBatchBytes}.");
        if (options.MaxOplogRows <= 0)
            failures.Add($"LocalDb:Replication:MaxOplogRows must be greater than 0; got {options.MaxOplogRows}.");
        if (options.MaxOplogAge <= TimeSpan.Zero)
@@ -38,6 +40,10 @@ public sealed class ReplicationOptionsValidator : IValidateOptions<ReplicationOp
            failures.Add($"LocalDb:Replication:ReconnectBackoffMax must be greater than zero; got {options.ReconnectBackoffMax}.");
        else if (options.ReconnectBackoffMax > ReconnectBackoffCeiling)
            failures.Add($"LocalDb:Replication:ReconnectBackoffMax must be at most {ReconnectBackoffCeiling}; got {options.ReconnectBackoffMax}.");
+        if (options.KeepAlivePingDelay <= TimeSpan.Zero)
+            failures.Add($"LocalDb:Replication:KeepAlivePingDelay must be greater than zero; got {options.KeepAlivePingDelay}.");
+        if (options.KeepAlivePingTimeout <= TimeSpan.Zero)
+            failures.Add($"LocalDb:Replication:KeepAlivePingTimeout must be greater than zero; got {options.KeepAlivePingTimeout}.");
        if (options.MaintenanceInterval <= TimeSpan.Zero)
            failures.Add($"LocalDb:Replication:MaintenanceInterval must be greater than zero; got {options.MaintenanceInterval}.");
        if (options.MaxHlcDriftAhead <= TimeSpan.Zero)
--- a/Show More
+++ b/Show More