[Bugfix] Disable vllm compilation cache due to compilation issues#95
Merged
tyler-griggs merged 3 commits intoNovaSky-AI:mainfrom Jul 17, 2025
Merged
[Bugfix] Disable vllm compilation cache due to compilation issues#95tyler-griggs merged 3 commits intoNovaSky-AI:mainfrom
tyler-griggs merged 3 commits intoNovaSky-AI:mainfrom
Conversation
tyler-griggs
approved these changes
Jul 17, 2025
SumanthRH
added a commit
that referenced
this pull request
Jul 17, 2025
fannie1208
pushed a commit
to vinid/SkyRL
that referenced
this pull request
Aug 19, 2025
…vaSky-AI#95) # What does this PR do? Disables vLLM compilation cache : https://docs.vllm.ai/en/v0.9.0/design/v1/torch_compile.html#compilation-cache We've seen issues especially in multi-node settings where the vLLM initialization fails for version >= 0.9.0 at compilation time: [compilation_error_multinode.log](https://github.com/user-attachments/files/21282993/compilation_error_multinode.log) The error has been hard to track down - since in single-node settings things have worked fine. Removing the vllm cache directory `~/.cache/vllm/` completely and retrying also doesn't fix the issue. There's likely a bug with cache write or so in vLLM. For now, we disable compilation cache with `VLLM_DISABLE_COMPILE_CACHE` . --------- Signed-off-by: SumanthRH <sumanthrh99@gmail.com>
fannie1208
pushed a commit
to vinid/SkyRL
that referenced
this pull request
Aug 19, 2025
Signed-off-by: SumanthRH <sumanthrh99@gmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Disables vLLM compilation cache : https://docs.vllm.ai/en/v0.9.0/design/v1/torch_compile.html#compilation-cache
We've seen issues especially in multi-node settings where the vLLM initialization fails for version >= 0.9.0 at compilation time:
compilation_error_multinode.log
The error has been hard to track down - since in single-node settings things have worked fine. Removing the vllm cache directory
~/.cache/vllm/completely and retrying also doesn't fix the issue. There's likely a bug with cache write or so in vLLM.For now, we disable compilation cache with
VLLM_DISABLE_COMPILE_CACHE.