Phase 0 of the ClusterClient→gRPC migration
(docs/plans/2026-07-22-clusterclient-to-grpc-plan.md). Standalone hardening: it
closes a gap that exists today and is a precondition for moving command/control
onto gRPC in later phases.
T0.1 — delete the ManagementActor ClusterClientReceptionist registration.
It was built for an out-of-cluster CLI that was never written: the shipped CLI
speaks HTTP Basic to /management, which asks the actor in-process through
ManagementActorHolder. Nothing in the repo ever sent to /user/management. The
actor still runs there; only the cross-boundary advertisement is gone. Six
documents claimed the CLI used ClusterClient — including the CLI's own README
"Architecture Notes" — and are corrected here rather than left to rot.
T0.2 — record, do not port, the dead integration-routing path.
IntegrationCallRequest is unwired at BOTH ends: RouteIntegrationCallAsync has
zero callers anywhere, and RegisterLocalHandler(Integration, …) appears only in
a test, so production always answers "Integration handler not available". It is
excluded from the gRPC contract (28 of 29 commands migrate) rather than
enshrined on an additive-only wire format, and deleting it during a
transport migration would mix a behavioural change into a change whose whole
value is that behaviour is identical. See
docs/known-issues/2026-07-22-integration-call-routing-is-dead-code.md.
T0.3 — preshared-key authentication on SiteStreamService.
The service shipped with no auth at all: plaintext h2c, no interceptor, so
anything that could reach a site node's :8083 could open a live data stream or
read audit rows back via PullAuditEvents/PullSiteCalls. ControlPlaneAuthInterceptor
now gates /sitestream.SiteStreamService/ — modeled on LocalDbSyncAuthInterceptor
(constant-time compare, fail-closed, PermissionDenied) but gating a SET of
service prefixes so phases 1A/1B add services rather than interceptors. LocalDb
sync keeps its own separate key: it authenticates the pair partner, not central,
and collapsing the two would make a site's central-facing key also admit writes
into its database.
Keys are per site (SB-GRPC-PSK-<siteId>), never fleet-wide, so a compromised
site yields only its own. Central attaches them through ControlPlaneCredentials,
which binds CallCredentials to the channel — covering unary and streaming
uniformly, and letting the key resolve asynchronously, which a client
interceptor could not do without blocking. All three central→site channel
creation sites go through it (SiteStreamGrpcClient and both audit pull invokers);
the pull invokers' channel caches are re-keyed by (site, endpoint) because
credentials are per-site and bound to the channel.
Two decisions beyond the plan:
* StartupValidator now requires GrpcPsk on Site nodes. The plan specified only
the runtime gate, but fail-closed with no boot check produces a node that
joins, answers heartbeats and reports healthy while refusing every stream,
audit pull and telemetry ingest — silent and total. Same reasoning as the
existing inbound API-key pepper rule.
* Added Communication:SitePsks as a central-side key map. The plan assumed
central would read the store, seeded via a dev KEK; the docker rig
deliberately boots with no master key, so store-only resolution would leave
it unable to dial its own sites. The store stays primary — it is the only
source that can serve a site added at runtime — with the map covering
key-less hosts and one-off pins. Neither source falling back to
"unauthenticated" is the invariant.
T0.4 — dev keys on both rigs and tests.
34 tests. The seven that matter most exercise a real in-process gRPC stack over
TestServer: the unit tests on either side of the wire would both stay green if
the halves disagreed, and gRPC refuses call credentials on a plaintext channel
by default — the UnsafeUseInsecureChannelCallCredentials opt-in is only provable
by making a real call. They confirm correct key passes on unary AND streaming,
wrong key and no-credentials both get PermissionDenied, and an unresolvable key
fails the call with nothing reaching the service.
OPERATIONAL: a site node upgraded to this build without a key will not boot.
That includes the gitignored deploy/wonder-app-vd03/ overlay.
Two-node keep-oldest could NEVER survive a crash of the oldest/active node:
Akka.NET 1.5.62 KeepOldest.OldestDecision only lets down-if-alone rescue a
side with >= 2 members, so the 1-vs-1 survivor takes DownReachable and downs
ITSELF — proven live on the rig ('SBR took decision ... including myself')
before this change. static-quorum(1) is worse (IsTooManyMembers -> DownAll);
keep-majority just re-keys the fatal crash to the lowest address.
SplitBrainResolverStrategy gains 'auto-down' (new default): BuildHocon emits
Akka's AutoDowning provider with auto-down-unreachable-after = StableAfter.
The leader among the REACHABLE members downs the unreachable peer, so the
survivor takes over singletons and /health/active in ~25s regardless of which
node died. Accepted trade (explicit owner decision): a real network partition
runs dual-active until an operator restarts one side. keep-oldest remains
supported; DownIfAlone validation is now scoped to it.
Live drill on the rebuilt rig: active-crash TAKEOVER in 28s (victim still
down; all 7 singletons Younger->Oldest), standby-crash removal 27s with 0
routing blips; victims rejoin as standby in 2s. New real-cluster tests pin
both directions (SbrFailoverTests.AutoDown_*); TwoNodeClusterFixture gains a
strategy knob. All 16 appsettings flipped (src, docker, docker-env2, and the
gitignored wonder-app-vd03 overlay on disk — owner must sync to the host).
Docs: decision record docs/plans/2026-07-21-auto-down-availability-decision.md,
Component-ClusterInfrastructure downing section rewritten, drill + README
reworked (active mode now asserts takeover), deferred-work SBR row resolved.
Tasks 14, 15 and 16, landed as ONE commit.
PLAN DEFECT: these three tasks cannot compile separately. SiteReplicationActor
takes a ReplicationService and calls ReplaceAllAsync (Task 14 deletes both);
DeploymentManagerActor Tells message types declared in ReplicationMessages.cs
(Task 15 deletes it); AkkaHostedService constructs the actor (Task 16). Any
ordering leaves a broken intermediate. Combining them also strengthens the
invariant Task 14 already stated for itself — the two mechanisms never both
run, and never neither.
Registered 8 tables in SiteLocalDbSetup.OnReady: sf_messages plus the 7 site
config tables. notification_lists and smtp_configurations are deliberately NOT
registered — permanently empty by design, so registering them would open a
standing replication channel whose only historical payload was plaintext SMTP
passwords. Migrate stays the LAST call in OnReady, after all registrations, so
migrated rows enter the oplog through live capture triggers.
Deleted: SiteReplicationActor, ReplicationMessages.cs, ReplicationService,
StoreAndForwardStorage.ReplaceAllAsync, and 6 test files. ReplaceAllAsync is
not merely unused but unsafe to keep: a mass DELETE on a now-replicated table
would be captured and shipped to the peer.
Kept ActiveNodeEvaluator (delivery gate + heartbeat still need it) with its doc
corrected, and activeNodeCheck in AkkaHostedService (SiteCommunicationActor).
The positional-argument hazard the plan flagged was real: removing
DeploymentManagerActor's optional IActorRef? replicationActor shifted 6
trailing optionals, and 4 test call sites bound the wrong arguments with no
compile error at some positions. Converted them to named arguments where
possible — Props.Create builds an expression tree, which rejects out-of-position
named args, so the rest are padded positionally with a comment saying why.
The Task 7 'not yet registered' test was INVERTED rather than deleted, and is
exact in both directions: too few means a table silently stops replicating, too
many means the SMTP tables leak. Added a separate security-named test for those
two, and a composite-PK test (LWW keys on the full PK, so a truncated key set
would collapse distinct rows). The convergence suites now get their
registrations from the real OnReady — their temporary harness registration is
deleted, so they prove the cutover rather than agreeing with themselves.
Verified: build 0 warnings; SiteRuntime 512, StoreAndForward 130, Host 330,
AuditLog 355, ExternalSystemGateway 142, HealthMonitoring 97, LocalDb
integration 16 — all pass, 0 failures.
Claude-Session: https://claude.ai/code/session_01BL2Vu1ESDQ9SCN4gVKkdts
Adds the active-central-node, in-memory, reference-counted per-site live alarm
cache backing the operator Alarm Summary page. No persisted central alarm store
([PERM]) — no EF entity/table/migration/DbSet.
- ISiteAlarmLiveCache (new): singleton seam — Subscribe(siteId, onChanged) ->
IDisposable (ref-counted, linger stop on last-out), GetCurrentAlarms, IsLive.
- SiteAlarmAggregatorActor (new): one per site. Seed-then-stream ordering copied
from DebugStreamBridgeActor — open the site-wide alarm-only gRPC stream first,
buffer live deltas while the snapshot fan-out runs, flush with per-key dedup
(AlarmKey = InstanceUniqueName|AlarmName|SourceReference), then live pass-through
into an in-memory dict. Placeholder rows seeded from the snapshot and never
expected on the stream; a real-alarm delta (distinct key) never wipes them.
NodeA<->NodeB reconnect (retry budget + stability window), reconnect re-seeds,
periodic reconcile (authoritative clear-and-rebuild) corrects instance-set drift,
and a budget-exhausted stream self-heals on the next reconcile tick.
- SiteAlarmLiveCacheService (new): DI singleton facade — viewer reference-counting,
linger-delayed last-out stop (version + TryRemove(ref) race guards), the
snapshot fan-out seed (RequestDebugSnapshotAsync per Enabled instance, capped),
bounded start-retry self-heal on transient start failure, and the immutable
published-snapshot store the page reads. Cache mutated only on the actor thread;
viewer callbacks invoked outside the lock.
- CommunicationOptions: LiveAlarmCacheLinger (30s), LiveAlarmCacheReconcileInterval
(60s), LiveAlarmCacheSeedConcurrency (8), LiveAlarmCacheMaxSubscribersPerSite
(200). Task 6 formalizes eager validation + telemetry.
- DI registration + AkkaHostedService SetActorSystem wiring on the active central node.
- Tests: 14 actor (seed/stream ordering, dedup, native-alarm parity, placeholder
coherence, reconnect re-seed, reconcile replace, self-heal, stop) + 6 service
(shared start, linger stop, resubscribe cancels stop, idempotent dispose, unknown
site, transient-start self-heal). Code-reviewer pass: no persistence, no Critical;
both Important findings addressed.
Claude-Session: https://claude.ai/code/session_01MtdgwpEeCUn6cUA5f1LMPj
WhenTerminated watchdog calls IHostApplicationLifetime.StopApplication() when
the ActorSystem dies outside StopAsync (SBR self-down + run-coordinated-shutdown-when-down),
so the service supervisor restarts the node as a fresh incarnation. Optional
ctor param keeps every existing construction site compiling.
Plan's deploy/wonder-app-vd03/install.ps1 service-recovery-actions edit is
skipped: that production deploy artifact is not tracked in this repo (Task 16
skipped its appsettings.Central.json for the same reason).
Remove project bookkeeping citations from shipped code comments across the
solution: hyphenated task IDs (WP-14, StoreAndForward-025), milestone/task/
issue refs (M3, Task 4, Audit Log #23, #21), Bundle X task-bundle labels,
and C/D/K/S/T phase labels.
Comment text only — no code logic, string/log literals, or XML-doc structure
changed. Genuine descriptions are preserved (only the citation is stripped),
and technical lookalikes are retained (UTF-8, SHA-256, T00:00:00, M365,
UTC-5, pre-C4/pre-C5 schema versions). Flagged by the new CommentChecker
TaskReferenceInComment / TrackingReferenceInComment checks plus targeted
grep passes; full solution builds clean, append-only guard tests pass.
Notify-and-fetch follow-ups:
- PendingDeploymentPurgeActor: a central cluster singleton (not
readiness-gated, best-effort) that sweeps expired PendingDeployment
staging rows on CommunicationOptions.PendingDeploymentPurgeInterval
(default 1h). Modeled on the kpi-history-recorder pattern: self-scheduling
timer, per-tick DI scope -> IDeploymentManagerRepository, continue-on-error.
Wired in AkkaHostedService.RegisterCentralActors (manager + proxy + drain);
resolves the deferred TODO in DeploymentService. Correctness never depends
on it (supersession bounds rows to <=1/instance; the fetch endpoint enforces
the TTL), so it is deliberately absent from RequiredSingletonsHealthCheck.
- SQL Server integration test for StagePendingIfAbsentAsync re-staging an
instance's OWN DeploymentId over an expired row against the real UNIQUE
index on DeploymentId — confirms EF orders DELETE before INSERT in one
SaveChanges (SQLite's constraint timing differs from SQL Server's). Plus
a same-instance supersession variant on real SQL Server.
Tests: 2 TestKit actor tests + 2 SQL Server integration tests (both ran
green against the infra MSSQL container); 235 Communication + 15
PendingDeployment tests pass; Host builds 0 warnings.
Wire the M6 KPI History recorder into the central composition path:
- Program.cs: call services.AddKpiHistory(configuration) on the central-only
branch alongside AddNotificationOutbox/AddAuditLog/AddSiteCallAudit.
- AkkaHostedService.cs: register KpiHistoryRecorderActor as a central,
non-role-scoped ClusterSingletonManager + ClusterSingletonProxy + a
PhaseClusterLeave CoordinatedShutdown graceful-stop drain (singleton name
'kpi-history-recorder'), copied/adapted from the audit-log-purge block.
- appsettings.Central.json (Host + docker + docker-env2 central nodes): add a
ScadaBridge:KpiHistory section (SampleInterval 00:01:00, RetentionDays 90,
PurgeInterval 1.00:00:00, DefaultMaxSeriesPoints 200).
KPI history is observability/best-effort and MUST NOT gate readiness: the
recorder is deliberately NOT added to RequiredSingletonsHealthCheck or any
other readiness gate.
Site Call Audit (#22): build the documented periodic reconciliation PULL
self-heal path for the eventually-consistent central SiteCalls mirror, as a
dedicated PullSiteCalls gRPC RPC kept separate from the audit pull. This is the
pull PLUMBING only; the central reconciliation tick is a separate follow-up.
- IOperationTrackingStore.ReadChangedSinceAsync(sinceUtc, batchSize): inclusive
UpdatedAtUtc cursor, oldest-first, batch-capped; SQLite impl projects tracking
rows onto SiteCallOperational (Kind->Channel, TargetSummary->Target, SourceSite
left empty - the store has no site-id column).
- sitestream.proto: rpc PullSiteCalls + PullSiteCallsRequest/Response, mirroring
PullAuditEvents; regenerated checked-in SiteStreamGrpc/*.cs.
- SiteCallDtoMapper.ToDto(SiteCallOperational): inverse of FromDto for the handler.
- SiteStreamGrpcServer.PullSiteCalls handler + SetOperationTrackingStore seam;
Host wires the seam alongside SetSiteAuditQueue (site roles only).
- Central IPullSiteCallsClient + GrpcPullSiteCallsClient (home: AuditLog/Central to
reuse ISiteEnumerator; SiteCallAudit does not reference AuditLog). Re-stamps
SourceSite from the dialed siteId; no-throw on tolerable transport faults;
SpecifyKind (not ToUniversalTime) cursor handling. Central-only DI registration.
Tests: ReadChangedSinceAsync (4), PullSiteCalls handler (6), GrpcPullSiteCallsClient
(8). Full solution build 0 warnings/0 errors (TreatWarningsAsErrors).
Per-probe health-check child scopes were disposing the AddTransient-bridged
ActorSystem (IDisposable), terminating the live cluster node ~4s after boot and
leaving every singleton-proxy Ask to hang the full 30s QueryTimeout — the central
report pages (/notifications, /site-calls, /monitoring/health) loaded in ~30s.
Bridge it as a singleton via a new lazy AkkaHostedService.GetOrCreateActorSystem()
so child-scope disposal never touches it. Verified: 0 post-startup terminates,
healthy active/standby, report pages ~0.05s, Playwright 68 passed / 0 failed.
Joseph Doherty