Merge pull request #697 from instructlab/dependabot/github_actions/py… #375
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
| # SPDX-License-Identifier: Apache-2.0 | |
| name: Functional GPU (NVIDIA Tesla T4 x1) | |
| on: | |
| workflow_dispatch: {} | |
| # run against every merge commit to 'main' and release branches | |
| push: | |
| branches: | |
| - main | |
| - release-* | |
| # only run on PRs that touch certain regex paths | |
| pull_request_target: | |
| branches: | |
| - main | |
| - release-* | |
| paths: | |
| # note this should match the merging criteria in 'mergify.yml' | |
| - "**.py" | |
| - "pyproject.toml" | |
| - "requirements**.txt" | |
| - "constraints-dev.txt" | |
| - "tox.ini" | |
| - ".github/workflows/functional-gpu-nvidia-t4-x1.yml" # This workflow | |
| concurrency: | |
| group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }} | |
| cancel-in-progress: true | |
| env: | |
| LC_ALL: en_US.UTF-8 | |
| defaults: | |
| run: | |
| shell: bash | |
| permissions: | |
| contents: read | |
| jobs: | |
| start-small-ec2-runner: | |
| runs-on: ubuntu-latest | |
| outputs: | |
| label: ${{ steps.launch-ec2-instance-with-fallback.outputs.label }} | |
| ec2-instance-id: ${{ steps.launch-ec2-instance-with-fallback.outputs.ec2-instance-id }} | |
| ec2-instance-region: ${{ steps.launch-ec2-instance-with-fallback.outputs.ec2-instance-region }} | |
| steps: | |
| - name: Checkout "launch-ec2-runner-with-fallback" in-house CI action | |
| uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0 | |
| with: | |
| repository: instructlab/ci-actions | |
| # clone the "ci-actions" repo to a local directory called "ci-actions", instead of | |
| # overwriting the current WORKDIR contents | |
| path: ci-actions | |
| ref: release-v0.1 | |
| sparse-checkout: | | |
| actions/launch-ec2-runner-with-fallback | |
| - name: Launch EC2 Runner with Fallback | |
| id: launch-ec2-instance-with-fallback | |
| uses: ./ci-actions/actions/launch-ec2-runner-with-fallback | |
| env: | |
| TMPDIR: "/tmp" | |
| with: | |
| aws_access_key_id: ${{ secrets.AWS_ACCESS_KEY_ID }} | |
| aws_secret_access_key: ${{ secrets.AWS_SECRET_ACCESS_KEY }} | |
| github_token: ${{ secrets.GH_PERSONAL_ACCESS_TOKEN }} | |
| regions_config: > | |
| [ | |
| { | |
| "region": "us-east-2", | |
| "subnets": { | |
| "us-east-2a": "${{ vars.SUBNET_US_EAST_2A }}", | |
| "us-east-2b": "${{ vars.SUBNET_US_EAST_2B }}", | |
| "us-east-2c": "${{ vars.SUBNET_US_EAST_2C }}" | |
| }, | |
| "ec2-ami": "${{ vars.AWS_EC2_AMI_US_EAST_2 }}", | |
| "security-group-id": "${{ vars.SECURITY_GROUP_ID_US_EAST_2 }}" | |
| }, | |
| { | |
| "region": "us-east-1", | |
| "subnets": { | |
| "us-east-1a": "${{ vars.SUBNET_US_EAST_1A }}", | |
| "us-east-1b": "${{ vars.SUBNET_US_EAST_1B }}", | |
| "us-east-1c": "${{ vars.SUBNET_US_EAST_1C }}", | |
| "us-east-1d": "${{ vars.SUBNET_US_EAST_1D }}", | |
| "us-east-1e": "${{ vars.SUBNET_US_EAST_1E }}", | |
| "us-east-1f": "${{ vars.SUBNET_US_EAST_1F }}" | |
| }, | |
| "ec2-ami": "${{ vars.AWS_EC2_AMI_US_EAST_1 }}", | |
| "security-group-id": "${{ vars.SECURITY_GROUP_ID_US_EAST_1 }}" | |
| } | |
| ] | |
| try_spot_instance_first: false | |
| ec2_instance_type: g4dn.2xlarge | |
| aws_resource_tags: > | |
| [ | |
| {"Key": "Name", "Value": "instructlab-ci-github-small-runner"}, | |
| {"Key": "GitHubRepository", "Value": "${{ github.repository }}"}, | |
| {"Key": "GitHubRef", "Value": "${{ github.ref }}"}, | |
| {"Key": "GitHubPR", "Value": "${{ github.event.number }}"} | |
| ] | |
| functional-gpu-small-test: | |
| needs: | |
| - start-small-ec2-runner | |
| runs-on: ${{ needs.start-small-ec2-runner.outputs.label }} | |
| # It is important that this job has no write permissions and has | |
| # no access to any secrets. This part is where we are running | |
| # untrusted code from PRs. | |
| permissions: {} | |
| steps: | |
| - name: Install Packages | |
| run: | | |
| cat /etc/os-release | |
| sudo dnf install -y gcc gcc-c++ make git python3.11 python3.11-devel | |
| - name: Checkout instructlab/sdg | |
| uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0 | |
| with: | |
| # https://github.com/actions/checkout/issues/249 | |
| fetch-depth: 0 | |
| - name: Fetch and checkout PR | |
| if: github.event_name == 'pull_request_target' | |
| run: | | |
| git fetch origin pull/${{ github.event.pull_request.number }}/merge:pr-merge-${{ github.event.pull_request.number }} | |
| git checkout pr-merge-${{ github.event.pull_request.number }} | |
| git log -1 --format="%H %s" | |
| - name: Install instructlab/sdg | |
| run: | | |
| export PATH="/home/ec2-user/.local/bin:/usr/local/cuda/bin:$PATH" | |
| python3.11 -m venv --upgrade-deps venv | |
| . venv/bin/activate | |
| nvidia-smi | |
| python3.11 -m pip install tox tox-gh>=1.2 -c constraints-dev.txt | |
| python3.11 -m pip cache remove llama_cpp_python | |
| CMAKE_ARGS="-DLLAMA_CUDA=on" python3.11 -m pip install -r requirements-dev.txt -c constraints-dev.txt | |
| - name: Check disk before tests | |
| run: | | |
| df -h | |
| df -i | |
| - name: Run functional gpu tests with tox | |
| run: | | |
| . venv/bin/activate | |
| tox -e functional-gpu | |
| - name: Check disk after tests | |
| if: ${{ always() }} | |
| run: | | |
| df -h | |
| df -i | |
| stop-small-ec2-runner: | |
| needs: | |
| - start-small-ec2-runner | |
| - functional-gpu-small-test | |
| runs-on: ubuntu-latest | |
| if: ${{ always() }} | |
| steps: | |
| - name: Configure AWS credentials | |
| uses: aws-actions/configure-aws-credentials@7474bc4690e29a8392af63c5b98e7449536d5c3a # v4.3.1 | |
| with: | |
| aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }} | |
| aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }} | |
| aws-region: ${{ needs.start-small-ec2-runner.outputs.ec2-instance-region }} | |
| - name: Stop EC2 runner | |
| uses: machulav/ec2-github-runner@fb91019e71385fb10dfcbec812b4de8c61589f7b # v2.4.1 | |
| with: | |
| mode: stop | |
| github-token: ${{ secrets.GH_PERSONAL_ACCESS_TOKEN }} | |
| label: ${{ needs.start-small-ec2-runner.outputs.label }} | |
| ec2-instance-id: ${{ needs.start-small-ec2-runner.outputs.ec2-instance-id }} | |
| functional-gpu-small-workflow-complete: | |
| # we don't want to block PRs on failed EC2 cleanup | |
| # so not requiring "stop-small-ec2-runner" as well | |
| needs: ["start-small-ec2-runner", "functional-gpu-small-test"] | |
| runs-on: ubuntu-latest | |
| steps: | |
| - name: Functional GPU Workflow Complete | |
| run: echo "Functional GPU Workflow Complete" |