Worker proc VllmWorker-1 died unexpectedly
Open
Nobody has claimed this yet.
bug
- Dominant language
- Python
- Stars
- 10.8k
- Forks
- 989
- Avg merge
- 6h 29m
- Merged PRs (30d)
- 85
Description
Already spoken a little with @bradhilton about this one! Popping here just in case others experience it too and for keeping track.
Happens only for multi-gpu.
Using Qwen/Qwen3-0.6B on 2x A100s (80GB)
On the first step after rewards are calculated, I get this:
DEBUG 07-04 17:35:10 [shm_broadcast.py:456] No available shared memory broadcast block found in 60 second.
ERROR 07-04 17:35:16 [multiproc_executor.py:140] Worker proc VllmWorker-1 died unexpectedly, shutting down executor.
DEBUG 07-04 17:36:10 [shm_broadcast.py:456] No available shared memory broadcast block found in 60 second.
DEBUG 07-04 17:37:10 [shm_broadcast.py:456] No available shared memory broadcast block found in 60 second.
DEBUG 07-04 17:38:10 [shm_broadcast.py:456] No available shared memory broadcast block found in 60 second.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with the reported two-GPU setup using Qwen/Qwen3-0.6B and review the messages from shm_broadcast.py and multiproc_executor.py. Reproduce the failure after rewards are calculated, then determine why the VllmWorker-1 process dies and shared-memory blocks remain unavailable. Done means the multi-GPU run no longer fails, or the issue is narrowed to a confirmed cause with supporting evidence.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- distributed-systems, machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 30/100