llm_server/routes/queue.py

@@ -36,6 +36,8 @@ class RedisPriorityQueue:
             print(f'Rejecting request from {item[1]} - {ip_count} requests in progress.')
             return None  # reject the request
 
+        print('--->', event.event_id)
+
         self.redis.zadd('queue', {json.dumps((item, event.event_id, selected_model)): -priority})
         self.increment_ip_count(item[1], 'queued_ip_count')
         return event

llm_server/workers/inferencer.py

@@ -33,6 +33,8 @@ def worker():
         increment_ip_count(client_ip, 'processing_ips')
         incr_active_workers(selected_model, backend_url)
 
+        print('<---', event_id.event_id)
+
         if not request_json_body:
             # This was a dummy request from the websocket handlers.
             # We're going to let the websocket handler decrement