merge: startup reconcile starts the healed InstanceActor (register row 67) — EnsureInstanceStartedCommand to the DM singleton, idempotent, disable-wins, Warning site event on self-heal

2026-08-15 17:56:21 -04:00
parent e8a90f18dc abf864b4e5
commit 744ff36a7b
7 changed files with 631 additions and 9 deletions
@@ -215,6 +215,8 @@ Both `AlarmActor` and `NativeAlarmActor` tell the `InstanceActor` an `AlarmState

 Config replication has no actor. `SiteReplicationActor` — which received `ReplicateConfigDeploy` / `ReplicateConfigRemove` / `ReplicateConfigSetEnabled` / `ReplicateArtifacts` / `ReplicateStoreAndForward` from the active node's `DeploymentManagerActor`, tracked the peer through cluster membership events, and forwarded each command to `/user/site-replication` via `ActorSelection` — was deleted in LocalDb Phase 2, together with its notify-and-fetch exchange (the standby was told a deploy had happened and then HTTP-fetched the config itself). The site's config tables are now replicated by LocalDb CDC, so a deploy on either node reaches the other as an ordinary row change. `SiteReconciliationActor` still fetches over HTTP at node startup when central reports gaps; that is a different path and it survives.

+Since register row 67 that pass also heals the **runtime**, not only the file. It runs after `DeploymentManagerActor`'s boot-time SQLite load, so a config it stored used to start no instance actor — an enabled instance missing at boot needed a *second* restart to run. Each enabled instance it heals now gets a fire-and-forget `EnsureInstanceStartedCommand` sent through the deployment-manager singleton proxy; the handler creates the instance actor only if one is not already running, re-reads the stored row to honour `is_enabled`, and steps aside for any in-flight deploy. A start emits a Warning log line and a `deployment` site event so the heal is visible.
+
 ## Usage

 ### Lifecycle commands