Repository navigation
fix(local_vllm_model): restore vLLM runtime deps the shared policy never-installs - #6183
qiaochuz-nv wants to merge 1 commit into
Conversation
…ver-installs #4837 copied the root never-install list into the vLLM servers' locks, so the venv Gym builds for local_vllm_model and genrm_model lacks pycountry, torchvision, torchaudio, torchcodec, opencv-python-headless and pynvvideocodec. vLLM 0.29 requires all six, and `vllm serve` fails at startup for every model: MistralCommonBackend needs pycountry, and the engine core imports torchvision. Re-enable them for these two components through scripts/component-lock-policy.json, the way toolsandbox (pycountry) and vlm_eval_kit/osworld (torchvision, opencv) do, using vLLM 0.29.0's own requirement strings under its non-Darwin marker. Add a test that every vLLM requirement is installed and the failing import chain loads. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: qiaochuz <qiaochuz@nvidia.com>
|
/ok to test 1cb323a |
svcnvidia-nemo-ci
left a comment
There was a problem hiding this comment.
SHIP
LGTM — no concerns found.
I reviewed all seven changed files and traced the managed-policy merge, component installation, model startup consumers, and test collection. The policy exceptions match both component declarations and locks without changing the root exclusions. The new tests check installed dependencies and the two reported failing imports in each component environment.
This was a static review. I didn't run dependency installation, tests, or GPU startup.
Note: formal approval is not available for this request, so this review is posted as a comment only.
Review generated by codex. This was a full review. Request a full review with /review reset=true.
|
Closing without merging. Leaving |
What does this PR do?
Restores the six vLLM runtime dependencies that the managed locks of
local_vllm_modelandgenrm_modelnever install. Without them,vllm servefails at startup for every model, text-only ones included.#4837 copied the root never-install list (
sys_platform == 'never') into these two servers'override-dependencies. Since then, the venv Gym builds for them lackspycountry,torchvision,torchaudio,torchcodec,opencv-python-headlessandpynvvideocodec. vLLM 0.29.0 lists the five codec packages in its ownRequires-Dist, andpycountrycomes throughmistral_common→pydantic-extra-types[pycountry]. Two imports break startup:transformers.MistralCommonBackendneedspycountry, so the API server dies before it loads a model;pycountryadded by hand, the engine core then dies onfrom torchvision.transforms import InterpolationMode(vllm/transformers_utils/processors/minimax_m3.py). It imports this even forQwen/Qwen3-0.6B.The fix follows the policy described in the root
pyproject.toml: "Servers that genuinely need these declare them in their own … override-dependencies". It records the exception inscripts/component-lock-policy.jsonfor both components, the way toolsandbox (pycountry), vlm_eval_kit and osworld (torchvision, opencv) and agentic_mme_agent (opencv) already do:sys_platform != 'darwin'marker as thevllm==0.29.0pin. Darwin (vllm 0.11) resolves exactly as before.pycountryis unpinned, as in toolsandbox.scripts/component_locks.pysynced the twopyproject.tomlfiles and relocked both. In eachuv.lock, only these six entries change, from the never-marker to the pins, plus their wheels. No other package version moves.tests/test_dependencies.pyin each server checks that every vLLM requirement is installed and imports the two modules that failed.Licensing: this reverses the codec and copyleft exclusion for these two server venvs only. The image still prefetches no server venv by default (
docker/Dockerfile,NEMO_GYM_PREFETCH_CONFIGS), anddocker/install_excluded_deps.shstays as is. Please confirm this exception is acceptable for the two vLLM servers; vLLM cannot serve without it.Validation
All on an H100 (Linux x86_64), uv 0.13.8. Each run builds a throwaway worktree with
uv sync --frozenin each component, thenvllm servefrom thelocal_vllm_modelvenv with an OpenAI-compatible request.Before: main
efd41fa161.The new test run against this venv:
2 failed(assert ['opencv-pyth...Codec==2.0.4'] == [],ModuleNotFoundError: No module named 'torchvision').After: this branch.
Also:
python scripts/component_locks.py --jobs 8: 301 projects checked, 0 failed.pre-commit runon the changed files: clean.efd41fa161all 57 single-GPU QA cases that startlocal_vllm_modelexited within about 20 s with this error.Rollouts
N/A: dependency-only change, with no server, agent or environment code changes. The failure happens at vLLM startup, before any rollout. The two
vllm serverequests above cover the server start.Compatibility and benchmark impact
The
local_vllm_modelandgenrm_modelvenvs are larger: they now install torchvision, torchaudio, torchcodec, opencv-python-headless, PyNvVideoCodec and pycountry. Darwin is unchanged. No benchmark behaviour changes, except that local-vLLM benchmarks start again.Checklist
pyproject.tomlalready describes this exception path.