Re-queue jobs when no workers are available instead of failing

vidai/cluster_master.py

@@ -1022,7 +1022,9 @@ class ClusterMaster:
                         else:
                             print(f"Failed to assign job {job['id']} to worker {worker_key}")
                     else:
-                        print(f"No suitable worker found for job {job['id']}")
+                        print(f"No suitable worker found for job {job['id']}, re-queuing")
+                        from .database import update_queue_status
+                        update_queue_status(job['id'], 'queued', error='No suitable worker found, re-queued')
 
                 await asyncio.sleep(5)  # Poll every 5 seconds
 

vidai/queue.py

@@ -262,8 +262,8 @@ class QueueManager:
         response = handle_web_message(message)
 
         if response and response.msg_type == 'error':
-            # Immediate error
-            update_queue_status(job['id'], 'failed', error=response.data.get('error', 'Backend error'))
+            # Immediate error - re-queue the job instead of failing
+            update_queue_status(job['id'], 'queued', error=response.data.get('error', 'No workers available, re-queued'))
             return
 
         # Poll for result (backend handles async processing)