Skip to content

improve NVIDIA CI stability #75

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
functionstackx merged 3 commits into from
Oct 1, 2025
Merged

improve NVIDIA CI stability #75

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
d1dc8c4
improve CI stability with docker
kedarpotdar-nv Sep 30, 2025
1d19374
add comment on docker init
kedarpotdar-nv Sep 30, 2025
d825e07
add comments
kedarpotdar-nv Oct 1, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion runners/launch_b200-nv.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@ srun --jobid=$JOB_ID bash -c "enroot import -o $SQUASH_FILE docker://$IMAGE"
srun --jobid=$JOB_ID \
--container-image=$SQUASH_FILE \
--container-mounts=$GITHUB_WORKSPACE:/workspace/,$HF_HUB_CACHE_MOUNT:$HF_HUB_CACHE \
--container-mount-home \
--no-container-mount-home --container-writable \
--container-workdir=/workspace/ \
--no-container-entrypoint --export=ALL \
bash benchmarks/${MODEL_CODE}_${PRECISION}_b200${FRAMEWORK_SUFFIX}_slurm.sh
Expand Down
12 changes: 10 additions & 2 deletions runners/launch_b200-nvd.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,13 +19,21 @@ if nvidia-smi --query-compute-apps=pid --format=csv,noheader | grep -q '[0-9]';
fi

set -x
docker run --rm -d --network host --name $server_name \
# Use --init flag to run an init process (PID 1) inside container for better signal handling and zombie process cleanup
# Ref: https://www.paolomainardi.com/posts/docker-run-init/

# NCCL_GRAPH_REGISTER tries to automatically enable user buffer registration with CUDA Graphs.
# Disabling it can reduce perf but will improve CI stability. i.e. we won't see vLLM/Sglang crashes.
# Ref: https://docs.nvidia.com/deeplearning/nccl/user-guide/docs/env.html#nccl-graph-register


docker run --rm -d --init --network host --name $server_name \
--runtime nvidia --gpus all --ipc host --privileged --shm-size=16g --ulimit memlock=-1 --ulimit stack=67108864 \
-v $HF_HUB_CACHE_MOUNT:$HF_HUB_CACHE \
-v $CACHE_DIR:/workspace/flashinfer_cache \
-v $GITHUB_WORKSPACE:/workspace/ -w /workspace/ \
-e HF_TOKEN -e HF_HUB_CACHE -e MODEL -e TP -e CONC -e MAX_MODEL_LEN -e ISL -e OSL -e PORT=$PORT \
-e FLASHINFER_WORKSPACE_BASE=/workspace/flashinfer_cache \
-e FLASHINFER_WORKSPACE_BASE=/workspace/flashinfer_cache -e NCCL_GRAPH_REGISTER=0 \
-e TORCH_CUDA_ARCH_LIST="10.0" -e CUDA_DEVICE_ORDER=PCI_BUS_ID -e CUDA_VISIBLE_DEVICES="0,1,2,3,4,5,6,7" \
--entrypoint=/bin/bash \
$(echo "$IMAGE" | sed 's/#/\//') \
Expand Down