G-8 (KEK rotation) — built in ZB.MOM.WW.Secrets, lib 0.1.2->0.1.3:
- ISecretCipher.Rewrap(row, oldKek, newKek): re-wraps the per-secret DEK only
(bodies never re-encrypted; revision/timestamps preserved -> invisible to
cluster LWW). Fail-closed on wrong old-KEK id, wrong key bytes, and malformed
wraps; DEK zeroed on all paths.
- ISecretStore.ApplyRewrapAsync(rewrappedRow, expectedCurrentWrappedDek):
updates only the 4 wrap columns + kek_id, compare-and-swap on the current
wrapped DEK so a concurrent set/rotate cannot corrupt a row (closes a
review-caught TOCTOU).
- KekRotationService.RewrapAllAsync + RewrapReport: enumerate all rows incl.
tombstones, idempotent/resumable skip-already-current, bounded CAS-retry,
fail-closed on unknown/identical KEK.
- `secret rewrap-all` CLI verb: key material only via env-var name / file path,
JSON counts report; README section + operator runbook.
Verified: full offline suite green (82 core + 15 UI, 0 regressions) + end-to-end
CLI smoke + adversarial crypto review (all 7 categories PASS; TOCTOU fixed).
G-7 (clustered replication) — designed + planned, no code:
- Fork resolved to build Option A (shared SQL-Server ISecretStore); Akka
replicator ZB.MOM.WW.Secrets.Akka is a deferred phase-2. Design doc +
executable plan + .tasks.json under docs/plans/2026-07-17-secrets-g7-*.
Tracking: components/secrets/GAPS.md + CLAUDE.md secrets row updated.
The v3.0 dual-namespace rewrite (PR #472) moved anchors and reshaped the
build, so the plan is refreshed to stay executable: CPM correction (OtOpcUa
is on Central Package Management, not inline versions), OpcUaClient options
in .Contracts + GalaxyDriverBrowser in .Browser, DriverHostActor line
shifts, and appsettings/owner precision. Driver-secret flow itself was
untouched by v3.0, so the two-layer approach holds.
Phase 0 of the shared-GLAuth standardization. config.toml = merged dc=zb,dc=local
directory (15 groups in partitioned 55xx/56xx/57xx families, 14 users incl.
multi-role spanning all groups, serviceaccount search account). compose runs one
glauth/glauth:latest on :3893. README is the deploy/verify runbook. Code-reviewed;
fixed scp -r idempotency in the deploy command (README + plan Task 4).
Approved design: consolidate OtOpcUa, MxAccessGateway, ScadaBridge dev/test auth
onto one shared GLAuth at 10.100.0.35:3893 (dc=zb,dc=local, plaintext). App-neutral
source of truth in scadaproj/infra/glauth/; merged directory with gid families
partitioned 55xx/56xx/57xx + multi-role/admin/serviceaccount; per-app Server
repoints; incremental rollout keeping old glauths until verified.
Per-app cutover steps mapped to the library surface; flags 5 findings that
change the plan (OtOpcUa section is Security:Ldap not Authentication:Ldap;
singleton 'bug' already mitigated; ScadaBridge inbound API keys are a
re-architecture not a reformat; OtOpcUa config+DB mapping + DevStubMode +
2nd LDAP consumer; MxGateway ApiKeys is the low-risk donor path).
Approved brainstorming output: two-library program (publish + adopt
ZB.MOM.WW.Auth then ZB.MOM.WW.Audit across OtOpcUa, MxAccessGateway,
ScadaBridge), library-major waterfall, ending with audit Actor wired
from the Auth principal. Local-only delivery; verified feed/source state.
Configuration is now adopted across all three sister apps (local branches),
so flip the status lines in CLAUDE.md, components/configuration/GAPS.md, and the
lib README/CLAUDE.md from 'not adopted' to adopted (also corrects 27->42 tests).
Adds the brainstorm design doc + bite-sized implementation plan (+tasks.json)
under docs/plans/ that drove the adoption.
13 tasks: Task 0 publishes/verifies the 2 nupkgs on Gitea (gates all); then 3
independent per-repo phases — OtOpcUa (1-3), ScadaBridge (4-6), MxGateway (7-11,
incl. the high-risk MEL->Serilog swap) — and Task 12 scadaproj bookkeeping last.
Records two behaviour-preserving refinements vs the design: ScadaBridge keeps
LoggerConfigurationFactory (+TraceContextEnricher) instead of AddZbSerilog, and
MxGateway keeps GatewayLogScope as-is. Breaking items #6/#7 deferred.
Second cross-fleet shared-library adoption (after Health). Full scope:
AddZbTelemetry (OTel Resource identity triple + standard instrumentation +
Prometheus /metrics) on all 3, plus shared Serilog on all 3 — including the
MxGateway MEL->Serilog migration. Records the correction that MxGateway's
logging was NOT actually adopted on main despite the docs' claim. Behaviour-
preserving bar; breaking items (#6 unit, #7 rename) deferred.
Detailed task-by-task plan (publish to Gitea, then per-repo behaviour-preserving
probe swaps) incorporating recon findings that revised the design: MxGateway worker
IPC is named pipes (custom SQLite readiness probe instead of gRPC), ScadaBridge
ActorSystem is not in DI (transient bridge), downstream gRPC probes + IDbContextFactory
switch + ScadaBridge seam unification deferred.
Plan to integrate the built-but-unadopted Health library into OtOpcUa,
MxAccessGateway, and ScadaBridge: Gitea-registry distribution, per-repo
behaviour-preserving probe swaps (preset-based), canonical tiers + writer,
MxGateway-first sequencing.
Joseph Doherty