Fix client state not being reset after failed Start()

When Start() fails due to a real error (e.g., database connection failure),
the client's internal state was left in a running state, preventing subsequent
Start() calls from succeeding.

Add StartFailed() method to BaseStartStop that properly resets internal state
after a startup failure. This is separate from Stop() handling - when Stop()
cancels the context (ErrStop), Stop() itself handles cleanup via finalizeStop().

Fixes the issue where a client could not be restarted after a transient
startup failure.

Author: Maciej Mucha

client.go

@@ -1071,15 +1071,15 @@ func (c *Client[TTx]) Start(ctx context.Context) error {
 
 		return nil
 	}(); err != nil {
-		defer func() {
-			stopped()
-			// Reset the client's internal state (isRunning flag) so that Start
-			// can be called again after a startup failure.
-			c.baseStartStop.Stop()
-		}()
+		// If context was cancelled due to Stop(), just close the stopped channel
+		// and return. Stop() will handle cleanup via finalizeStop().
 		if errors.Is(context.Cause(fetchCtx), startstop.ErrStop) {
+			stopped()
 			return nil
 		}
+
+		// For real startup failures, reset state so Start() can be called again.
+		c.baseStartStop.StartFailed(stopped)
 		return err
 	}
 

rivershared/startstop/start_stop.go

@@ -136,6 +136,28 @@ func (s *BaseStartStop) StartInit(ctx context.Context) (context.Context, bool, f
 	}
 }
 
+// StartFailed should be called when a service fails to start after StartInit.
+// It closes the stopped channel and resets internal state so Start can be
+// called again.
+//
+// This should not be used when a Stop is already in progress (ErrStop), because
+// Stop will handle cleanup via finalizeStop.
+func (s *BaseStartStop) StartFailed(stopped func()) {
+	if s.cancelFunc != nil {
+		s.cancelFunc(ErrStop)
+	}
+	stopped()
+
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if !s.isRunning {
+		return
+	}
+	s.isRunning = false
+	s.started = nil
+	s.stopped = nil
+}
+
 // Started returns a channel that's closed when a service finishes starting, or
 // if failed to start and is stopped instead. It can be used in conjunction with
 // WaitAllStarted to verify startup of a constellation of services.