fix(communication): malformed site gRPC address keeps the old channel, never aborts the refresh tick or restarts CentralCommunicationActor (review 2026-10-21 03/R-1)

2026-10-22 06:14:34 -04:00
parent 0fd9bcbce6
commit b0a97a4b67
5 changed files with 261 additions and 8 deletions
@@ -392,8 +392,19 @@ public class CentralCommunicationActor : ReceiveActor
    private void HandleSiteAddressCacheLoaded(SiteAddressCacheLoaded msg)
    {
        // Per-site transport resource reconciliation (build/drop gRPC channel
-        // pairs). Runs on the actor thread each refresh tick.
-        _transport.ReconcileSites(msg);
+        // pairs). Runs on the actor thread each refresh tick. Guarded (review 2026-10-21 03/R-1):
+        // a throw here would restart this actor, whose PreStart re-reads the Sites table at
+        // TimeSpan.Zero — a restart loop for as long as a bad row persists — and would skip the
+        // aggregator prune below.
+        try
+        {
+            _transport.ReconcileSites(msg);
+        }
+        catch (Exception ex)
+        {
+            _log.Error(ex, "Reconciling site gRPC channels from the refreshed address cache failed; "
+                + "existing channels are kept until the next refresh");
+        }

        // Self-healing eviction: a site deleted from configuration would otherwise
        // linger in the aggregator as a permanently-offline tile (and live KPI