fix(webapp): reschedule reconnect when subscribe() returns stopped

LogicalReplicationClient.subscribe() can resolve without throwing or
emitting an "error" event when leader-lock acquisition fails — it just
calls this.stop() and returns. The reconnect callback now checks
isStopped after subscribe() and throws so the recovery handler can
schedule the next attempt instead of silently giving up.

Author: ericallam

apps/webapp/app/services/runsReplicationService.server.ts

@@ -264,6 +264,13 @@ export class RunsReplicationService {
       logger: this.logger,
       reconnect: async () => {
         await this._replicationClient.subscribe(this._latestCommitEndLsn ?? undefined);
+        if (this._replicationClient.isStopped) {
+          // subscribe() can resolve without throwing or emitting an "error"
+          // event when leader-lock acquisition fails (see LogicalReplication-
+          // Client.subscribe leader-election branch). Throw here so the
+          // recovery handler reschedules the next attempt.
+          throw new Error("Replication client stopped after subscribe()");
+        }
       },
       isShuttingDown: () => this._isShuttingDown || this._isShutDownComplete,
     });

apps/webapp/app/services/sessionsReplicationService.server.ts

@@ -245,6 +245,11 @@ export class SessionsReplicationService {
       logger: this.logger,
       reconnect: async () => {
         await this._replicationClient.subscribe(this._latestCommitEndLsn ?? undefined);
+        if (this._replicationClient.isStopped) {
+          // See RunsReplicationService for the rationale: subscribe() can
+          // resolve without throwing when leader-lock acquisition fails.
+          throw new Error("Replication client stopped after subscribe()");
+        }
       },
       isShuttingDown: () => this._isShuttingDown || this._isShutDownComplete,
     });