AgentOps: [OPENAI WRAPPER] Error in async_chat_completion_stream_wrapper: Error code: 503

[450586509]

error comes from this directory: examples/calc_x

Processing request of type ListToolsRequest Processing request of type ListToolsRequest Processing request of type ListToolsRequest Processing request of type ListToolsRequest 127.0.0.1 - - [14/Aug/2025 16:48:58] "POST /traces HTTP/1.1" 200 - 127.0.0.1 - - [14/Aug/2025 16:48:58] "POST /traces HTTP/1.1" 200 - 127.0.0.1 - - [14/Aug/2025 16:48:58] "POST /traces HTTP/1.1" 200 - 127.0.0.1 - - [14/Aug/2025 16:48:58] "POST /traces HTTP/1.1" 200 - 🖇 AgentOps: [OPENAI WRAPPER] Error in async_chat_completion_stream_wrapper: Error code: 503 🖇 AgentOps: [OPENAI WRAPPER] Error in async_chat_completion_stream_wrapper: Error code: 503 🖇 AgentOps: [OPENAI WRAPPER] Error in async_chat_completion_stream_wrapper: Error code: 503 🖇 AgentOps: [OPENAI WRAPPER] Error in async_chat_completion_stream_wrapper: Error code: 503 Failure: Error code: 503 Failure: Error code: 503 answer: None ground_truth: 8 reward: 0.0 Failure: Error code: 503 Failure: Error code: 503 answer: None ground_truth: 20 reward: 0.0 answer: None ground_truth: 1/10 reward: 0.0 answer: None ground_truth: 4 reward: 0.0

2025-08-14 16:49:43,232 [INFO] (Process-29480 agentlightning.runner) [Worker 1 | Rollout rollout-afd29e05-8ce3-43d5-8481-47dcc046b8bf] Completed in 46.04s. Reward: 0.0 2025-08-14 16:49:43,232 [INFO] (Process-29481 agentlightning.runner) [Worker 2 | Rollout rollout-687e1a88-fd17-4268-a1f6-6a31f29c2a41] Completed in 46.02s. Reward: 0.0 2025-08-14 16:49:43,233 [INFO] (Process-29482 agentlightning.runner) [Worker 3 | Rollout rollout-1504611f-3f85-4c0a-8a0b-49a5473f04ec] Completed in 46.04s. Reward: 0.0 2025-08-14 16:49:43,237 [INFO] (Process-29480 agentlightning.client) [Task 3 Received] ID: rollout-10ac3339-a9a5-410d-a704-03773b02c556

#printed by me openai_base_url=http://127.0.0.1:41981/v1 model=Qwen/Qwen2.5-1.5B-Instruct

[ultmaster]

Command to run and steps to reproduce the problems please.

[450586509]

I followed this example: Calc-X Example In one terminal, run:

bash ../../scripts/restart_ray.sh python calc_agent.py

In another terminal, run:

bash train.sh

[ultmaster]

what's the logs of train.sh?

[450586509]

calc_agent.log train.log

calc_agent.log is generated by the following commands:

bash ../../scripts/restart_ray.sh python calc_agent.py

train.log is generated by the following command:

bash train.sh

[RENEK-bool]

I also encountered this same problem

[ultmaster]

answer: None ground_truth: 5 reward: 0.0 You can see no answer is derived. This is an issue with the agent itself.

Debug with autogen under verl is difficult. You may try this first.