mirror of
https://github.com/microsoft/mscclpp.git
synced 2026-05-12 01:10:22 +00:00
## Summary - **Multi-node H100 CI setup**: Improve architecture detection and GPU configuration - **Remove hardcoded VMSS hostnames** from deploy files - **Fix CUDA compat library issue**: Remove stale compat paths from Docker image for CUDA 12+. Instead, `peer_access_test` now returns a distinct exit code (2) for CUDA init failure, and `setup.sh` conditionally adds compat libs only when needed. This fixes `cudaErrorSystemNotReady` (error 803) when the host driver is newer than the container's compat libs. - **Speed up deploy**: Replace recursive `parallel-scp` with tar+scp+untar to avoid per-file SSH overhead. --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
86 lines
3.5 KiB
Bash
86 lines
3.5 KiB
Bash
set -ex
|
|
|
|
TEST_NAME=$1
|
|
IB_ENVIRONMENT="${2:-true}"
|
|
PLATFORM="${3:-cuda}"
|
|
|
|
KeyFilePath=${SSHKEYFILE_SECUREFILEPATH}
|
|
ROOT_DIR="${SYSTEM_DEFAULTWORKINGDIRECTORY}/"
|
|
if [ "${TEST_NAME}" == "nccltest-single-node" ]; then
|
|
ROOT_DIR="${ROOT_DIR}/mscclpp"
|
|
SYSTEM_DEFAULTWORKINGDIRECTORY="${SYSTEM_DEFAULTWORKINGDIRECTORY}/mscclpp"
|
|
fi
|
|
DST_DIR="/tmp/mscclpp"
|
|
if [ "${TEST_NAME}" == "nccltest-single-node" ] || [ "${TEST_NAME}" == "single-node-test" ]; then
|
|
HOSTFILE="${SYSTEM_DEFAULTWORKINGDIRECTORY}/test/deploy/hostfile_ci"
|
|
else
|
|
HOSTFILE="${SYSTEM_DEFAULTWORKINGDIRECTORY}/test/deploy/hostfile"
|
|
fi
|
|
SSH_OPTION="StrictHostKeyChecking=no"
|
|
|
|
chmod 400 ${KeyFilePath}
|
|
ssh-keygen -t rsa -f sshkey -P ""
|
|
|
|
while true; do
|
|
set +e
|
|
parallel-ssh -i -t 0 -h ${HOSTFILE} -x "-i ${KeyFilePath}" -O $SSH_OPTION "hostname"
|
|
if [ $? -eq 0 ]; then
|
|
break
|
|
fi
|
|
echo "Waiting for sshd to start..."
|
|
sleep 5
|
|
done
|
|
|
|
set -e
|
|
parallel-ssh -i -t 0 -h ${HOSTFILE} -x "-i ${KeyFilePath}" -O $SSH_OPTION "sudo rm -rf ${DST_DIR}"
|
|
tar czf /tmp/mscclpp.tar.gz -C ${ROOT_DIR} .
|
|
parallel-scp -t 0 -h ${HOSTFILE} -x "-i ${KeyFilePath}" -O $SSH_OPTION /tmp/mscclpp.tar.gz /tmp/mscclpp.tar.gz
|
|
parallel-ssh -i -t 0 -h ${HOSTFILE} -x "-i ${KeyFilePath}" -O $SSH_OPTION \
|
|
"sudo mkdir -p ${DST_DIR} && sudo tar xzf /tmp/mscclpp.tar.gz -C ${DST_DIR} && sudo rm -f /tmp/mscclpp.tar.gz"
|
|
rm -f /tmp/mscclpp.tar.gz
|
|
|
|
if [ "${PLATFORM}" == "rocm" ]; then
|
|
parallel-ssh -i -t 0 -h ${HOSTFILE} -x "-i ${KeyFilePath}" -O $SSH_OPTION "sudo modprobe amdgpu"
|
|
fi
|
|
|
|
# Install GDRCopy kernel module on host VMs (CUDA only)
|
|
GDRCOPY_VERSION="2.5.2"
|
|
if [ "${PLATFORM}" == "cuda" ]; then
|
|
parallel-ssh -i -t 0 -h ${HOSTFILE} -x "-i ${KeyFilePath}" -O $SSH_OPTION \
|
|
"if lsmod | grep -q gdrdrv; then
|
|
echo 'gdrdrv module already loaded'
|
|
else
|
|
set -e
|
|
sudo apt-get update -y && sudo apt-get install -y build-essential devscripts debhelper check libsubunit-dev fakeroot pkg-config dkms
|
|
cd /tmp && wget -q https://github.com/NVIDIA/gdrcopy/archive/refs/tags/v${GDRCOPY_VERSION}.tar.gz -O gdrcopy.tar.gz
|
|
tar xzf gdrcopy.tar.gz && cd gdrcopy-${GDRCOPY_VERSION}/packages
|
|
CUDA=/usr/local/cuda ./build-deb-packages.sh
|
|
sudo dpkg -i gdrdrv-dkms_${GDRCOPY_VERSION}*.deb
|
|
sudo modprobe gdrdrv
|
|
rm -rf /tmp/gdrcopy.tar.gz /tmp/gdrcopy-${GDRCOPY_VERSION}
|
|
fi"
|
|
fi
|
|
|
|
# force to pull the latest image
|
|
parallel-ssh -i -t 0 -h ${HOSTFILE} -x "-i ${KeyFilePath}" -O $SSH_OPTION \
|
|
"sudo docker pull ${CONTAINERIMAGE}"
|
|
|
|
LAUNCH_OPTION="--gpus=all"
|
|
if [ "${PLATFORM}" == "rocm" ]; then
|
|
LAUNCH_OPTION="--device=/dev/kfd --device=/dev/dri --group-add=video"
|
|
fi
|
|
if [ "${IB_ENVIRONMENT}" == "true" ]; then
|
|
parallel-ssh -i -t 0 -h ${HOSTFILE} -x "-i ${KeyFilePath}" -O $SSH_OPTION \
|
|
"sudo docker run --rm -itd --privileged --net=host --ipc=host ${LAUNCH_OPTION} \
|
|
-w /root -v ${DST_DIR}:/root/mscclpp -v /opt/microsoft:/opt/microsoft --ulimit memlock=-1:-1 --name=mscclpp-test \
|
|
--entrypoint /bin/bash ${CONTAINERIMAGE}"
|
|
else
|
|
parallel-ssh -i -t 0 -h ${HOSTFILE} -x "-i ${KeyFilePath}" -O $SSH_OPTION \
|
|
"sudo docker run --rm -itd --net=host --ipc=host ${LAUNCH_OPTION} --cap-add=SYS_ADMIN --security-opt seccomp=unconfined \
|
|
-w /root -v ${DST_DIR}:/root/mscclpp -v /opt/microsoft:/opt/microsoft --ulimit memlock=-1:-1 --name=mscclpp-test \
|
|
--entrypoint /bin/bash ${CONTAINERIMAGE}"
|
|
fi
|
|
parallel-ssh -i -t 0 -h ${HOSTFILE} -x "-i ${KeyFilePath}" -O $SSH_OPTION \
|
|
"sudo docker exec -t --user root mscclpp-test bash '/root/mscclpp/test/deploy/setup.sh' ${PLATFORM}"
|
|
|