[BugFix] Fix the bug where non-DP0 processes fail to update file hotness in DP scenarios. #992

Workflow file for this run

.github/workflows/pull-request.yml at 57962f3

	name: 'Pull Request Gate'

	on:
	pull_request:
	branches: [ "dev", "main", "release", "feature*" ]

	jobs:
	pre-check:
	runs-on: ubuntu-latest
	steps:
	- uses: actions/checkout@v4
	with:
	fetch-depth: 0

	- id: fetch_base_and_pr_branches
	env:
	HEAD_REF: ${{ github.event.pull_request.head.ref }}
	run: \|
	git fetch origin ${{ github.base_ref }}

	if [ "${{ github.event.pull_request.head.repo.full_name }}" != "${{ github.repository }}" ]; then
	echo "PR is from a fork, adding remote."
	git remote add fork ${{ github.event.pull_request.head.repo.clone_url }}
	git fetch fork $HEAD_REF
	HEAD_BRANCH_NAME=fork/$HEAD_REF
	echo "HEAD_BRANCH_NAME=$HEAD_BRANCH_NAME" >> $GITHUB_OUTPUT
	else
	git fetch origin $HEAD_REF
	HEAD_BRANCH_NAME=origin/$HEAD_REF
	echo "HEAD_BRANCH_NAME=$HEAD_BRANCH_NAME" >> $GITHUB_OUTPUT
	fi

	# protect the workflows dir, only allow specific users to modify
	- id: permission-check
	env:
	ACTOR: ${{ github.actor }}
	run: \|
	CHANGED_FILES=$(git diff --name-only origin/${{ github.base_ref }}...${{ steps.fetch_base_and_pr_branches.outputs.HEAD_BRANCH_NAME }})
	echo "CHANGED_FILES=$CHANGED_FILES"

	if ! echo "$CHANGED_FILES" \| grep -q '^\.github/workflows/'; then
	echo "No .github/workflows changes, allow."
	exit 0
	fi

	ALLOWED_USERS=("dante159753" "mag1c-h" "yuanzhg078")

	echo ".github changes detected, check if user is allowed..."
	ACTOR="${{ github.actor }}"
	echo "PR author: $ACTOR"

	for u in "${ALLOWED_USERS[@]}"; do
	if [[ "$ACTOR" == "$u" ]]; then
	echo "Authorized user, allowing change."
	exit 0
	fi
	done

	echo "::error:: Only privileged users may modify .github/workflows/"
	exit 1

	- id: uptodate-check
	run: \|

	if git merge-base --is-ancestor "origin/${{ github.base_ref }}" "${{ steps.fetch_base_and_pr_branches.outputs.HEAD_BRANCH_NAME }}"; then
	echo "PR branch is up-to-date with target branch, continue workflow."
	else
	echo "::error::PR branch is OUT-OF-DATE with target branch, stop workflow, please update your branch and try again."
	exit 1
	fi

	lint-and-unit-tests:
	needs: pre-check
	uses: ./.github/workflows/lint-and-test.yml

	test-build-npu:
	timeout-minutes: 15
	runs-on: ubuntu-24.04-arm
	steps:
	- uses: actions/checkout@v4
	with:
	fetch-depth: 0

	- name: Set up Docker Buildx
	uses: docker/setup-buildx-action@v3

	- name: Build Image
	uses: docker/build-push-action@v5
	with:
	context: .
	file: ./docker/Dockerfile.vllm_npu
	build-args: "ENABLE_SPARSE=false"
	tags: ucm-npu:latest
	push: false
	load: false
	cache-from: type=gha,scope=npu
	cache-to: type=gha,mode=max,scope=npu

	test-build-npu-sparse:
	timeout-minutes: 15
	runs-on: ubuntu-24.04-arm
	steps:
	- uses: actions/checkout@v4
	with:
	fetch-depth: 0

	- name: Set up Docker Buildx
	uses: docker/setup-buildx-action@v3

	- name: Build Image
	uses: docker/build-push-action@v5
	with:
	context: .
	file: ./docker/Dockerfile.vllm_npu_v0110
	build-args: "ENABLE_SPARSE=true"
	tags: ucm-npu-sparse:latest
	push: false
	load: false
	cache-from: type=gha,scope=npu
	cache-to: type=gha,mode=max,scope=npu

	test-e2e-pc-gpu:
	timeout-minutes: 25
	runs-on: ["gpu-test-in-docker"]
	needs: lint-and-unit-tests
	permissions:
	checks: write
	pull-requests: write
	steps:
	- name: Clean repo
	run: \|
	if [ -d "${{github.workspace}}" ]; then
	cd ${{github.workspace}}
	rm -rf ./*
	rm -rf .[!.]*
	fi
	- uses: actions/checkout@v4
	- name: Install Docker CLI
	run: \|
	if ! command -v docker &> /dev/null; then
	echo "Docker CLI not found, installing..."
	sudo apt-get update
	sudo apt-get install -y docker.io
	else
	echo "Docker CLI already installed"
	fi
	- name: Generate Docker Image Version
	id: version
	run: \|
	DATE=$(date +%Y%m%d)
	SHORT_SHA=$(echo '${{ github.sha }}' \| cut -c1-7)
	REF_NAME=$(echo '${{ github.ref_name }}' \| tr '/' '-')
	VERSION="${REF_NAME}-${DATE}-${{ github.run_number }}-${SHORT_SHA}"
	echo "version=${VERSION}" >> $GITHUB_OUTPUT
	echo "Docker image version: ${VERSION}"
	- name: Build
	run: \|
	cd ${{github.workspace}}
	sudo -E docker build --network=host \
	--build-arg http_proxy="${http_proxy:-}" \
	--build-arg https_proxy="${https_proxy:-}" \
	--build-arg ENABLE_SPARSE=false \
	-t ucm-e2etest-online-inference:${{ steps.version.outputs.version }} \
	-f ./docker/Dockerfile.vllm_gpu ./
	- name: Test E2E Online Inference in Docker
	run: \|
	sudo chmod -R 777 /workspace/test_results/
	sudo docker run --rm \
	--gpus all \
	--ipc=host \
	-v /home/models:/home/models \
	-v /home/yanzhao/pipeline_results:/workspace/test_results \
	ucm-e2etest-online-inference:${{ steps.version.outputs.version }} \
	-c "cd /workspace/unified-cache-management/test && pip install -r requirements.txt && python3 -m pytest -x --stage=1 --feature=online_inference --junitxml=/workspace/test_results/${{ steps.version.outputs.version }}-online-inference.xml"
	- name: Upload pytest results
	uses: EnricoMi/publish-unit-test-result-action/linux@v2
	if: (!cancelled())
	with:
	files: \|
	/workspace/test_results/${{ steps.version.outputs.version }}-online-inference.xml
	check_name: Online inference test results
	- name: Cleanup Docker Image
	if: always()
	run: \|
	sudo docker rmi ucm-e2etest-online-inference:${{ steps.version.outputs.version }} \|\| true
	- name: Cleanup Test Results
	if: always()
	run: \|
	sudo rm -f /workspace/test_results/${{ steps.version.outputs.version }}-online-inference.xml \|\| true

	test-e2e-sparse-gpu:
	timeout-minutes: 25
	runs-on: ["gpu-test-in-docker"]
	needs: lint-and-unit-tests
	permissions:
	checks: write
	pull-requests: write
	steps:
	- name: Clean repo
	run: \|
	if [ -d "${{github.workspace}}" ]; then
	cd ${{github.workspace}}
	rm -rf ./*
	rm -rf .[!.]*
	fi
	- uses: actions/checkout@v4
	- name: Install Docker CLI
	run: \|
	if ! command -v docker &> /dev/null; then
	echo "Docker CLI not found, installing..."
	sudo apt-get update
	sudo apt-get install -y docker.io
	else
	echo "Docker CLI already installed"
	fi
	- name: Generate Docker Image Version
	id: version
	run: \|
	DATE=$(date +%Y%m%d)
	SHORT_SHA=$(echo '${{ github.sha }}' \| cut -c1-7)
	REF_NAME=$(echo '${{ github.ref_name }}' \| tr '/' '-')
	VERSION="${REF_NAME}-${DATE}-${{ github.run_number }}-${SHORT_SHA}"
	echo "version=${VERSION}" >> $GITHUB_OUTPUT
	echo "Docker image version: ${VERSION}"
	- name: Build
	run: \|
	cd ${{github.workspace}}
	sudo -E docker build --network=host \
	--build-arg http_proxy="${http_proxy:-}" \
	--build-arg https_proxy="${https_proxy:-}" \
	--build-arg ENABLE_SPARSE=true \
	-t ucm-e2etest-gpu-sparse:${{ steps.version.outputs.version }} \
	-f ./docker/Dockerfile.vllm_gpu_v0110 ./
	- name: Test E2E in Docker
	run: \|
	sudo chmod -R 777 /workspace/test_results/
	sudo docker run --rm \
	--gpus all \
	--ipc=host \
	-v /home/models:/home/models \
	-v /home/yanzhao/pipeline_results:/workspace/test_results \
	ucm-e2etest-gpu-sparse:${{ steps.version.outputs.version }} \
	-c "cd /workspace/unified-cache-management/test && pip install -r requirements.txt && python3 -m pytest -x --stage=1 --feature=online_inference_sparse --junitxml=/workspace/test_results/${{ steps.version.outputs.version }}-online-inference-sparse.xml"
	- name: Upload pytest results
	uses: EnricoMi/publish-unit-test-result-action/linux@v2
	if: (!cancelled())
	with:
	files: \|
	/workspace/test_results/${{ steps.version.outputs.version }}-online-inference-sparse.xml
	check_name: Sparse attention test results
	- name: Cleanup Docker Image
	if: always()
	run: \|
	sudo docker rmi ucm-e2etest-gpu-sparse:${{ steps.version.outputs.version }} \|\| true
	- name: Cleanup Test Results
	if: always()
	run: \|
	sudo rm -f /workspace/test_results/${{ steps.version.outputs.version }}-online-inference-sparse.xml \|\| true

	test-e2e-sparse-offline-gpu:
	runs-on: ["gpu-test-in-docker"]
	needs: lint-and-unit-tests
	permissions:
	checks: write
	pull-requests: write
	steps:
	- name: Clean repo
	run: \|
	if [ -d "${{github.workspace}}" ]; then
	cd ${{github.workspace}}
	rm -rf ./*
	rm -rf .[!.]*
	fi
	- uses: actions/checkout@v4
	- name: Install Docker CLI
	run: \|
	if ! command -v docker &> /dev/null; then
	echo "Docker CLI not found, installing..."
	sudo apt-get update
	sudo apt-get install -y docker.io
	else
	echo "Docker CLI already installed"
	fi
	- name: Generate Docker Image Version
	id: version
	run: \|
	DATE=$(date +%Y%m%d)
	SHORT_SHA=$(echo '${{ github.sha }}' \| cut -c1-7)
	REF_NAME=$(echo '${{ github.ref_name }}' \| tr '/' '-')
	VERSION="${REF_NAME}-${DATE}-${{ github.run_number }}-${SHORT_SHA}"
	echo "version=${VERSION}" >> $GITHUB_OUTPUT
	echo "Docker image version: ${VERSION}"
	- name: Build
	run: \|
	cd ${{github.workspace}}
	sudo -E docker build --network=host \
	--build-arg http_proxy="${http_proxy:-}" \
	--build-arg https_proxy="${https_proxy:-}" \
	--build-arg ENABLE_SPARSE=true \
	-t ucm-e2etest-gpu-sparse:${{ steps.version.outputs.version }} \
	-f ./docker/Dockerfile.vllm_gpu_v0110 ./
	- name: Test E2E in Docker
	run: \|
	sudo chmod -R 777 /workspace/test_results/
	sudo docker run --rm \
	--gpus all \
	--ipc=host \
	-v /home/models:/home/models \
	-v /home/yanzhao/pipeline_results:/workspace/test_results \
	ucm-e2etest-gpu-sparse:${{ steps.version.outputs.version }} \
	-c "cd /workspace/unified-cache-management/test && pip install -r requirements.txt && python3 -m pytest -x --stage=1 --feature=offline_inference_sparse --junitxml=/workspace/test_results/${{ steps.version.outputs.version }}-offline-inference-sparse.xml"
	- name: Upload pytest results
	uses: EnricoMi/publish-unit-test-result-action/linux@v2
	if: (!cancelled())
	with:
	files: \|
	/workspace/test_results/${{ steps.version.outputs.version }}-offline-inference-sparse.xml
	check_name: Sparse attention test results
	- name: Cleanup Docker Image
	if: always()
	run: \|
	sudo docker rmi ucm-e2etest-gpu-sparse:${{ steps.version.outputs.version }} \|\| true
	- name: Cleanup Test Results
	if: always()
	run: \|
	sudo rm -f /workspace/test_results/${{ steps.version.outputs.version }}-offline-inference-sparse.xml \|\| true

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[BugFix] Fix the bug where non-DP0 processes fail to update file hotness in DP scenarios. #992

Workflow file

[BugFix] Fix the bug where non-DP0 processes fail to update file hotness in DP scenarios. #992

Uh oh!

Workflow file for this run