Repository navigation
[Bugfix] Use dashes instead of underscores for --gpu-memory-utilization flag - #1124
Merged
ruizhang0101 merged 1 commit intoOct 7, 2026
Conversation
- Update helm/templates/deployment-vllm-multi.yaml - Update helm/tests/deployment-vllm-multi_test.yaml - Update operator/internal/controller/vllmruntime_controller.go Fixes inconsistency in CLI flag formatting across test files and controller code. Signed-off-by: antar-ghuman <antarghuman82@gmail.com>
antar-ghuman
requested review from
ApostaC,
YuhanLiu11 and
ruizhang0101
as code owners
October 6, 2026 20:56
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request updates the command-line argument for GPU memory utilization from --gpu_memory_utilization to --gpu-memory-utilization across the Helm templates, Helm tests, and the vLLM runtime controller to ensure consistency with the expected vLLM CLI argument format. There are no review comments, and I have no additional feedback to provide.
ruizhang0101
enabled auto-merge (squash)
October 7, 2026 18:24
This was referenced Oct 20, 2026
rahulssv-ibm
added a commit
to rahulssv-ibm/production-stack
that referenced
this pull request
Oct 20, 2026
The Helm chart and the operator pass `--max_loras` to `vllm serve`, while vLLM documents its engine args with dashes (`--max-loras`). The Ray cluster template already uses `--max-loras`. This is the same issue reported for `--gpu-memory-utilization` in vllm-project#1116; that flag was fixed in vllm-project#1124, but `--max_loras` was left behind. - helm: emit `--max-loras` in deployment-vllm-multi.yaml - operator: emit `--max-loras` in the VLLMRuntime deployment builder - add a helm unittest asserting `--max-loras` is rendered - add an operator unit test asserting the vLLM args use dashes Fixes vllm-project#1116 Signed-off-by: Rahul <Rahul.Vishwakarma2@ibm.com>
rahulssv-ibm
added a commit
to rahulssv-ibm/production-stack
that referenced
this pull request
Oct 20, 2026
The Helm chart and the operator pass `--max_loras` to `vllm serve`, while vLLM documents its engine args with dashes (`--max-loras`) and the Ray cluster template already uses `--max-loras`. vLLM's argument parser also accepts the underscore form, so this aligns the flag with the documented name rather than fixing a startup failure. This is the same issue reported for `--gpu-memory-utilization` in vllm-project#1116; that flag was fixed in vllm-project#1124, which left `--max_loras` behind. Fixes vllm-project#1116 Signed-off-by: Rahul <Rahul.Vishwakarma2@ibm.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.
Summary
Fixes inconsistency in CLI flag formatting by replacing underscores with dashes in the
--gpu-memory-utilizationflag across templates, test files, and controller code.Changes
helm/templates/deployment-vllm-multi.yamlhelm/tests/deployment-vllm-multi_test.yamloperator/internal/controller/vllmruntime_controller.goThis ensures the flag matches the correct CLI argument format and maintains consistency across the codebase.
Checklist
-swhen doinggit commit[Bugfix]