You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
System tray's Upstream Server count for active does not match what the submenu show #187
This was caused by a lock-ordering issue in the supervisor that let the stateview snapshot drift from the actual upstream manager state — under concurrent reconciliation, stateMu was held while calling methods that needed manager.mu, which dropped runtime events and froze the snapshot.
Fixed by:
Pre-fetching GetAllStates() and IsUserLoggedOut()before acquiring stateMu.Lock() in supervisor.reconcile()
Rebuilding updateSnapshot() from live manager states (not stale snapshot)
Lowercasing state strings in stateview so the health calculator normalises correctly
The system tray, web UI, and CLI all read from the same stateview snapshot now, so the count and submenu can no longer disagree.
Closing — please reopen if you still see drift on v0.29.2+.
Activity
Dumbris commentedon May 1, 2026
This was caused by a lock-ordering issue in the supervisor that let the stateview snapshot drift from the actual upstream manager state — under concurrent reconciliation,
stateMuwas held while calling methods that neededmanager.mu, which dropped runtime events and froze the snapshot.Fixed by:
GetAllStates()andIsUserLoggedOut()before acquiringstateMu.Lock()insupervisor.reconcile()updateSnapshot()from live manager states (not stale snapshot)The system tray, web UI, and CLI all read from the same stateview snapshot now, so the count and submenu can no longer disagree.
Closing — please reopen if you still see drift on v0.29.2+.