Add ping forwarding from backend to cluster_master to reset job timeouts

vidai/backend.py

@@ -197,6 +197,16 @@ def handle_worker_message(message: Message, client_sock) -> None:
             # Update progress timestamp to reset timeout
             job_progress_times[job_id] = time.time()
             print(f"PING received for job {job_id} - resetting timeout")
+            # Notify cluster master of ping to reset job timeout
+            try:
+                import socket
+                notification_sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+                notification_sock.connect(('localhost', 5004))  # Cluster master port + 1
+                notification_sock.sendall(f"ping:{job_id}".encode())
+                notification_sock.close()
+                print(f"Notified cluster master of ping for job {job_id}")
+            except Exception as e:
+                print(f"Failed to notify cluster master of ping: {e}")
     elif message.msg_type in ['analyze_response', 'train_response']:
         # Store result for web to poll
         pending_results[message.msg_id] = message

vidai/cluster_master.py

@@ -242,6 +242,11 @@ class ClusterMaster:
                 print("Received worker registration notification - checking for pending jobs")
                 # Trigger immediate job assignment check
                 await self._check_pending_jobs()
+            elif message.startswith("ping:"):
+                job_id = message.split(":", 1)[1]
+                if job_id in self.active_jobs:
+                    self.active_jobs[job_id]['last_progress'] = time.time()
+                    print(f"Ping received for job {job_id} - resetting timeout")
         except Exception as e:
             print(f"Notification handling error: {e}")
         finally: