fix(aorta): container /dev/kfd access and --override argument packing - #327
Open
speriaswamy-amd wants to merge 1 commit into
Open
fix(aorta): container /dev/kfd access and --override argument packing#327speriaswamy-amd wants to merge 1 commit into
speriaswamy-amd wants to merge 1 commit into
Conversation
Two defects on the existing single-node path, both surfaced by live cluster validation: - _launch_container() never passed `user`, so the container ran as the image's default UID (jenkins on the validation cluster) and could not open /dev/kfd despite --privileged. The `render` group was also missing. The function's own comment already claimed "Run as root so the container can access GPUs" - the code now matches it. - Training overrides were emitted as `--override k=v --override k2=v2`. Aorta's train.py declares `--override` with `nargs="*"`, so argparse keeps only the last group and every preceding override was silently dropped. All key=value tokens now sit behind a single `--override`. Extracts the env-building block out of run() into _build_base_env() so the override packing is unit-testable without a container. Co-Authored-By: Claude <noreply@anthropic.com>
This was referenced Aug 14, 2026
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Stack 2/6 — splits #171. Base: #326.
Why
Two defects on the existing single-node path, both surfaced by live 2-node validation but neither multi-node specific.
_launch_container()never passeduser, so the container ran as the image's default UID (jenkinson the validation cluster) and could not open/dev/kfddespite--privileged. Therendergroup was also missing. The function's own comment already claimed "Run as root so the container can access GPUs" — the code now matches it.--override k=v --override k2=v2. Aorta'strain.pydeclares--overridewithnargs="*", so argparse keeps only the last group — every preceding override was silently dropped.What changed
cvs/runners/aorta.py—user="root",group_add=["video", "render"]; allkey=valuetokens packed behind a single--override.run()into_build_base_env()so the override packing is unit-testable without a container.Test
ruffclean. Unit tests 589 → 595 (6 new: env construction, single---overrideinvariant, container user/groups).