GCP: Test Image #14
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
| name: "GCP: Test Image" | |
| on: | |
| workflow_dispatch: | |
| inputs: | |
| version_major: | |
| description: 'AlmaLinux major version' | |
| required: true | |
| default: '10' | |
| type: choice | |
| options: | |
| - 10-kitten | |
| - 10 | |
| - 9 | |
| - 8 | |
| arch: | |
| description: 'Architecture to test' | |
| required: true | |
| default: 'ALL' | |
| type: choice | |
| options: | |
| - ALL | |
| - x86_64 | |
| - aarch64 | |
| image_override: | |
| description: 'Image to test, overrides version_major to test a direct image instead. Architecture must be set properly for the image being passed. This must be a full path to a GCP image, for example, projects/almalinux-dev-images-469421/global/images/almalinux-9-v20230920' | |
| required: false | |
| default: '' | |
| x86_shape_override: | |
| description: 'x86_64 shape to use for jobs that do not iterate over matrix.shape (e.g. test-gcp-nonpershape). Empty = let cloud-image-tests choose. Example: c4-standard-8.' | |
| required: false | |
| default: '' | |
| aarch64_shape_override: | |
| description: 'aarch64 shape to use for jobs that do not iterate over matrix.shape (e.g. test-gcp-nonpershape). Empty = let cloud-image-tests choose. Example: c4a-standard-8.' | |
| required: false | |
| default: '' | |
| cit_git_repo: | |
| description: 'owner/repo of a cloud-image-tests fork to build instead of pulling the prebuilt image. Empty = use gcr.io/compute-image-tools/cloud-image-tests:latest. Full URLs are NOT accepted (actions/checkout takes owner/repo only). Example: GoogleCloudPlatform/cloud-image-tests' | |
| required: false | |
| default: '' | |
| cit_git_ref: | |
| description: 'Branch, tag, or commit SHA inside cit_git_repo. Empty = repository default branch. Ignored when cit_git_repo is empty.' | |
| required: false | |
| default: '' | |
| # notify_mattermost: | |
| # description: "Send notification to Mattermost" | |
| # required: true | |
| # type: boolean | |
| # default: false | |
| env: | |
| # Default zone fallback list for cit-run-with-retry. Multi-region spread so a | |
| # single-region capacity event doesn't fail the whole run. Shapes with | |
| # narrower availability override this via matrix `zones:` and/or `zone:`. | |
| GCP_DEFAULT_ZONES: "us-central1-a us-central1-b us-central1-c us-central1-f us-south1-a us-south1-b us-south1-c us-west1-a us-west1-b us-west1-c northamerica-northeast1-a northamerica-northeast1-b europe-west1-b europe-west1-c europe-west1-d europe-west2-a europe-west2-b europe-west2-c europe-west3-a europe-west3-b europe-west3-c europe-west4-a europe-west4-b europe-west4-c" | |
| jobs: | |
| init-data: | |
| runs-on: ubuntu-latest | |
| outputs: | |
| image_path: ${{ steps.determine_image.outputs.image_path }} | |
| steps: | |
| - name: Determine image to test | |
| id: determine_image | |
| run: | | |
| if [ -n "${{ inputs.image_override }}" ]; then | |
| echo "Using image override: ${{ inputs.image_override }}" | |
| image_path="${{ inputs.image_override }}" | |
| elif [ "${{ inputs.arch }}" = "ALL" ]; then | |
| echo "Using version major: ${{ inputs.version_major }}" | |
| echo "Using all architectures" | |
| image_path="projects/almalinux-dev-images-469421/global/images/family/almalinux-${{ inputs.version_major }}" | |
| else | |
| echo "Using version major: ${{ inputs.version_major }}" | |
| echo "Using arch: ${{ inputs.arch }}" | |
| image_path="projects/almalinux-dev-images-469421/global/images/family/almalinux-${{ inputs.version_major }}" | |
| fi | |
| echo "Determined image path: ${image_path}" | |
| echo "image_path=${image_path}" >> $GITHUB_OUTPUT | |
| # Optional: build cloud-image-tests from a user-supplied git repo+ref and | |
| # publish a docker tarball as an artifact. All test jobs `needs:` this; when | |
| # cit_git_repo is empty this job is skipped, which GitHub Actions treats as | |
| # non-blocking (skipped != failed), so test jobs run as today. | |
| build-cit: | |
| name: Build cloud-image-tests from source | |
| if: inputs.cit_git_repo != '' | |
| needs: init-data | |
| # family=m8azn + cpu=4 -> m8azn.xlarge (highest single-thread perf in AWS, | |
| # best for Go compile). Note: runs-on's inline labels do NOT support | |
| # `instance=<type>` — only `family=` + `cpu=`/`ram=`/etc. To pin a | |
| # different AWS size, change cpu (4=xlarge, 8=2xlarge, 16=4xlarge, ...). | |
| # volume=500g grows the actual root EBS volume (where /home/runner lives) | |
| # rather than attaching a secondary disk. Requires runs-on >= 2.9 — the | |
| # older `disk=` selector mounts an additional volume at /runs-on/ which | |
| # doesn't help when /home/runner/_diag/ fills up on the small fixed root. | |
| # Forks fall back to the default GH-hosted runner. Tune size based on | |
| # the runs-on cost/performance summary printed by the step below. | |
| runs-on: "${{ github.repository_owner == 'AlmaLinux' && format('runs-on={0}/family=m8azn/cpu=4/volume=500g/spot=false/image=ubuntu24-full-x64', github.run_id) || 'ubuntu-24.04' }}" | |
| steps: | |
| - name: Show runs-on cost/performance summary | |
| if: github.repository_owner == 'AlmaLinux' | |
| uses: runs-on/action@v1 | |
| - name: Checkout CIT source | |
| uses: actions/checkout@v6 | |
| with: | |
| repository: ${{ inputs.cit_git_repo }} | |
| ref: ${{ inputs.cit_git_ref }} | |
| path: cit-src | |
| - name: Build CIT image | |
| shell: bash | |
| run: docker build -t cit-custom:run-${{ github.run_id }} cit-src | |
| - name: Save image to tarball | |
| shell: bash | |
| run: docker save cit-custom:run-${{ github.run_id }} | gzip > cit-image.tar.gz | |
| - name: Upload CIT image artifact | |
| uses: actions/upload-artifact@v7 | |
| with: | |
| name: cit-custom-image | |
| path: cit-image.tar.gz | |
| retention-days: 1 | |
| # this runs an initial set of smoke tests against a shape that we can use with high concurrency | |
| # this is to speed up the feedback loop in case something obvious can be caught quickly with a new image | |
| # without having to wait on the full run of per-shape tests which takes a while | |
| test-gcp-initialtest: | |
| name: AlmaLinux ${{ inputs.image_override || format('{0} {1}', inputs.version_major, matrix.arch) }} Initial Tests | |
| needs: [init-data, build-cit] | |
| permissions: | |
| id-token: write | |
| contents: read | |
| runs-on: ubuntu-24.04 | |
| strategy: | |
| fail-fast: false | |
| matrix: | |
| # this sets the arch matrix based on the input | |
| # if input is ALL, then set to both x86_64 and aarch64 | |
| # otherwise set to the selected arch | |
| arch: ${{ fromJSON(inputs.arch == 'ALL' && '["x86_64","aarch64"]' || format('["{0}"]', inputs.arch)) }} | |
| steps: | |
| # we don't need the checked out files, but this is required for the google auth action to work | |
| - uses: 'actions/checkout@v6' | |
| - id: 'google-auth-image-testing' | |
| uses: 'google-github-actions/auth@v3' | |
| with: | |
| workload_identity_provider: 'projects/527193872801/locations/global/workloadIdentityPools/github-actions/providers/github' | |
| service_account: 'github-actions-image-testing@almalinux-image-testing-469421.iam.gserviceaccount.com' | |
| - name: 'Set up Google Cloud SDK' | |
| uses: 'google-github-actions/setup-gcloud@v3' | |
| - name: Download custom CIT image | |
| if: inputs.cit_git_repo != '' | |
| uses: actions/download-artifact@v7 | |
| with: | |
| name: cit-custom-image | |
| - name: Load custom CIT image | |
| if: inputs.cit_git_repo != '' | |
| shell: bash | |
| run: gunzip -c cit-image.tar.gz | docker load | |
| - name: 'Run Google cloud-image-testing tests which are hard-coded to specific shapes' | |
| uses: ./.github/actions/cit-run-with-retry | |
| with: | |
| image: "${{ needs.init-data.outputs.image_path }}${{ inputs.image_override == '' && matrix.arch == 'aarch64' && '-arm64' || '' }}" | |
| filter: '^(lssd|disk|vmspec)$' | |
| shape_flag: "${{ matrix.arch == 'aarch64' && '-arm64_shape c4a-standard-8' || '-x86_shape c4-standard-8' }}" | |
| parallel_count: '20' | |
| parallel_stagger: '1s' | |
| zones: ${{ env.GCP_DEFAULT_ZONES }} | |
| creds_path: ${{ env.GOOGLE_GHA_CREDS_PATH }} | |
| quota_log_file: ${{ runner.temp }}/quota-failures.jsonl | |
| image_ref: ${{ inputs.cit_git_repo != '' && format('cit-custom:run-{0}', github.run_id) || 'gcr.io/compute-image-tools/cloud-image-tests:latest' }} | |
| - name: Upload quota-failure log | |
| if: always() | |
| uses: actions/upload-artifact@v7 | |
| with: | |
| name: quota-failures-initialtest-${{ matrix.arch }} | |
| path: ${{ runner.temp }}/quota-failures.jsonl | |
| if-no-files-found: ignore | |
| retention-days: 7 | |
| # this runs tests which are hard-coded to specific shapes in CIT | |
| # running them against all shapes would be pointless as they are forced to specific shapes anyway | |
| test-gcp-nonpershape: | |
| name: AlmaLinux ${{ inputs.image_override || format('{0} {1}', inputs.version_major, matrix.arch) }} Non-Per-Shape Tests | |
| needs: [init-data, build-cit] | |
| permissions: | |
| id-token: write | |
| contents: read | |
| runs-on: ubuntu-24.04 | |
| strategy: | |
| fail-fast: false | |
| matrix: | |
| # this sets the arch matrix based on the input | |
| # if input is ALL, then set to both x86_64 and aarch64 | |
| # otherwise set to the selected arch | |
| arch: ${{ fromJSON(inputs.arch == 'ALL' && '["x86_64","aarch64"]' || format('["{0}"]', inputs.arch)) }} | |
| steps: | |
| # we don't need the checked out files, but this is required for the google auth action to work | |
| - uses: 'actions/checkout@v6' | |
| - id: 'google-auth-image-testing' | |
| uses: 'google-github-actions/auth@v3' | |
| with: | |
| workload_identity_provider: 'projects/527193872801/locations/global/workloadIdentityPools/github-actions/providers/github' | |
| service_account: 'github-actions-image-testing@almalinux-image-testing-469421.iam.gserviceaccount.com' | |
| - name: 'Set up Google Cloud SDK' | |
| uses: 'google-github-actions/setup-gcloud@v3' | |
| - name: Download custom CIT image | |
| if: inputs.cit_git_repo != '' | |
| uses: actions/download-artifact@v7 | |
| with: | |
| name: cit-custom-image | |
| - name: Load custom CIT image | |
| if: inputs.cit_git_repo != '' | |
| shell: bash | |
| run: gunzip -c cit-image.tar.gz | docker load | |
| - name: 'Run Google cloud-image-testing tests which are hard-coded to specific shapes' | |
| uses: ./.github/actions/cit-run-with-retry | |
| with: | |
| image: "${{ needs.init-data.outputs.image_path }}${{ inputs.image_override == '' && matrix.arch == 'aarch64' && '-arm64' || '' }}" | |
| filter: '^(lssd|disk|vmspec)$' | |
| # If an arch-matching shape_override input is set, pin the shape; | |
| # otherwise leave empty and let cloud-image-tests pick its hardcoded shapes. | |
| shape_flag: "${{ matrix.arch == 'aarch64' && inputs.aarch64_shape_override != '' && format('-arm64_shape {0}', inputs.aarch64_shape_override) || matrix.arch == 'x86_64' && inputs.x86_shape_override != '' && format('-x86_shape {0}', inputs.x86_shape_override) || '' }}" | |
| parallel_stagger: '10s' | |
| zones: ${{ env.GCP_DEFAULT_ZONES }} | |
| creds_path: ${{ env.GOOGLE_GHA_CREDS_PATH }} | |
| quota_log_file: ${{ runner.temp }}/quota-failures.jsonl | |
| image_ref: ${{ inputs.cit_git_repo != '' && format('cit-custom:run-{0}', github.run_id) || 'gcr.io/compute-image-tools/cloud-image-tests:latest' }} | |
| - name: Upload quota-failure log | |
| if: always() | |
| uses: actions/upload-artifact@v7 | |
| with: | |
| name: quota-failures-nonpershape-${{ matrix.arch }} | |
| path: ${{ runner.temp }}/quota-failures.jsonl | |
| if-no-files-found: ignore | |
| retention-days: 7 | |
| test-gcp-pershape-x86_64: | |
| name: ${{ inputs.image_override || format('{0}', inputs.version_major) }} x86_64 ${{ matrix.shape }} | |
| needs: [init-data, test-gcp-initialtest, build-cit] | |
| permissions: | |
| id-token: write | |
| contents: read | |
| runs-on: "${{ github.repository_owner == 'AlmaLinux' && format('runs-on={0}/runner=2cpu-linux-x64/spot=false/image=almalinux-10-x86_64', github.run_id) || 'ubuntu-24.04' }}" | |
| if: inputs.arch == 'ALL' || inputs.arch == 'x86_64' | |
| strategy: | |
| fail-fast: false | |
| matrix: | |
| shape: | |
| - n4-standard-2 | |
| - n4-standard-80 | |
| - n2-standard-4 | |
| - n2-standard-128 | |
| - n2d-standard-2 | |
| - n2d-standard-224 | |
| - n1-standard-1 | |
| - n1-standard-96 | |
| - c4-standard-2 | |
| # can never get capacity for 288 so using 192 | |
| - c4-standard-192 | |
| # all lssd shapes fail | |
| # https://github.com/GoogleCloudPlatform/cloud-image-tests/issues/345 | |
| # - c4-standard-4-lssd | |
| # - c4-standard-288-lssd | |
| # never has capacity anywhere | |
| # - c4-standard-288-metal | |
| - c4d-standard-2 | |
| # can never get capacity for 384 so using 192 | |
| - c4d-standard-192 | |
| # - c4d-standard-8-lssd | |
| # - c4d-standard-384-lssd | |
| - c3-standard-4 | |
| - c3-standard-176 | |
| # <failure message="" type="Failure"><![CDATA[step "create-vms" run error: googleapi: Error 400: C3 Metal instance does not support multi-vNIC., badRequest]]></failure> | |
| # - c3-standard-192-metal | |
| # lssd always fails | |
| # - c3-standard-4-lssd | |
| # - c3-standard-176-lssd | |
| - c3d-standard-4 | |
| - c3d-standard-360 | |
| # - c3d-standard-8-lssd | |
| # - c3d-standard-360-lssd | |
| - e2-standard-2 | |
| - e2-standard-32 | |
| - e2-medium | |
| - t2d-standard-1 | |
| - t2d-standard-60 | |
| # h4d tests fail | |
| # https://github.com/GoogleCloudPlatform/cloud-image-tests/issues/346 | |
| # - h4d-standard-192 | |
| # - h4d-highmem-192-lssd | |
| # can never get capacity for this shape | |
| # - h3-standard-88 | |
| - c2-standard-4 | |
| - c2-standard-60 | |
| - c2d-standard-2 | |
| - c2d-standard-112 | |
| # m4, x4, m3, m2 disabled due to no quotas | |
| # - m4-megamem-28 | |
| # - m4-megamem-224 | |
| # - x4-megamem-960-metal | |
| # - x4-megamem-1920-metal | |
| # - m3-megamem-64 | |
| # - m2-megamem-416 | |
| # m1 quota too low | |
| # - m1-megamem-96 | |
| # z3 tests fail | |
| # https://github.com/GoogleCloudPlatform/cloud-image-tests/issues/346 | |
| # z3 tests not needed, basically c3 with lots of lssd, so covered by c3-lssd tests | |
| # - z3-highmem-14-standardlssd | |
| # things with special zone requirements or other special needs | |
| include: | |
| # Placeholder entry that declares the optional `zone:` / `zones:` | |
| # keys on the matrix type so `matrix.zone` / `matrix.zones` | |
| # expressions in the cit-run-with-retry call are valid. Empty | |
| # values are no-ops at runtime (the action falls back to | |
| # env.GCP_DEFAULT_ZONES). Add real overrides below as needed. | |
| - shape: n4-standard-2 | |
| zone: '' | |
| zones: '' | |
| # only available in us-central1-b | |
| # disabled, never any capacity available | |
| # - shape: c4d-standard-384-metal | |
| # zone: us-central1-b | |
| # zones: us-central1-b | |
| steps: | |
| # we don't need the checked out files, but this is required for the google auth action to work | |
| - uses: 'actions/checkout@v6' | |
| - id: 'google-auth-image-testing' | |
| uses: 'google-github-actions/auth@v3' | |
| with: | |
| workload_identity_provider: 'projects/527193872801/locations/global/workloadIdentityPools/github-actions/providers/github' | |
| service_account: 'github-actions-image-testing@almalinux-image-testing-469421.iam.gserviceaccount.com' | |
| - name: 'Set up Google Cloud SDK' | |
| uses: 'google-github-actions/setup-gcloud@v3' | |
| - name: Runner OS specific configuration | |
| shell: bash | |
| run: | | |
| # Runner OS specific configuration | |
| if [ -e /etc/redhat-release ]; then | |
| sudo setenforce 0 | |
| sudo dnf -y install podman | |
| runner_user=$USER | |
| sudo loginctl enable-linger $runner_user | |
| docker_cmd=podman | |
| elif lsb_release -cs > /dev/null 2>&1; then | |
| docker_cmd=docker | |
| else | |
| echo "[Debug] Unknown OS" | |
| exit 1 | |
| fi | |
| echo "docker_cmd=${docker_cmd}" >> $GITHUB_ENV | |
| - name: Download custom CIT image | |
| if: inputs.cit_git_repo != '' | |
| uses: actions/download-artifact@v7 | |
| with: | |
| name: cit-custom-image | |
| - name: Load custom CIT image | |
| if: inputs.cit_git_repo != '' | |
| shell: bash | |
| # build-cit emits a docker-save tarball; podman 4.x+ reads it natively. | |
| run: gunzip -c cit-image.tar.gz | ${{ env.docker_cmd }} load | |
| - name: 'Run Google cloud-image-testing tests on ${{ matrix.shape }}' | |
| uses: ./.github/actions/cit-run-with-retry | |
| with: | |
| runtime: ${{ env.docker_cmd }} | |
| image: ${{ needs.init-data.outputs.image_path || inputs.image_override }} | |
| filter: '^(cvm|livemigrate|suspendresume|loadbalancer|guestagent|hostnamevalidation|imageboot|licensevalidation|network|security|hotattach|packagevalidation|ssh|metadata)$' | |
| shape_flag: "-x86_shape ${{ matrix.shape }}" | |
| parallel_count: '1' | |
| first_attempt_zone: ${{ matrix.zone || '' }} | |
| zones: ${{ matrix.zones || env.GCP_DEFAULT_ZONES }} | |
| creds_path: ${{ env.GOOGLE_GHA_CREDS_PATH }} | |
| quota_log_file: ${{ runner.temp }}/quota-failures.jsonl | |
| image_ref: ${{ inputs.cit_git_repo != '' && format('cit-custom:run-{0}', github.run_id) || 'gcr.io/compute-image-tools/cloud-image-tests:latest' }} | |
| - name: Upload quota-failure log | |
| if: always() | |
| uses: actions/upload-artifact@v7 | |
| with: | |
| name: quota-failures-pershape-x86_64-${{ matrix.shape }} | |
| path: ${{ runner.temp }}/quota-failures.jsonl | |
| if-no-files-found: ignore | |
| retention-days: 7 | |
| test-gcp-pershape-aarch64: | |
| name: ${{ inputs.image_override || format('{0}', inputs.version_major) }} aarch64 ${{ matrix.shape }} | |
| needs: [init-data, test-gcp-initialtest, build-cit] | |
| permissions: | |
| id-token: write | |
| contents: read | |
| runs-on: ubuntu-24.04 | |
| if: inputs.arch == 'ALL' || inputs.arch == 'aarch64' | |
| strategy: | |
| fail-fast: false | |
| matrix: | |
| shape: | |
| # pretty limited quotas for this one | |
| # let's test one shape flavor + lssd | |
| # metal gets large-CPUs detection check | |
| - c4a-standard-4 | |
| - c4a-standard-72 | |
| # all lssd shapes fail | |
| # https://github.com/GoogleCloudPlatform/cloud-image-tests/issues/345 | |
| #- c4a-standard-4-lssd | |
| # pretty limited quotas for this one | |
| # let's just test one shape flavor | |
| # can never get capacity for this shape, even with low parallelism, so skipping for now | |
| # - t2a-standard-4 | |
| - n4a-standard-4 | |
| include: | |
| # Placeholder so `parallel_count` / `zone` / `zones` exist on the | |
| # matrix type (the cit-run-with-retry call below references them | |
| # via `matrix.X || <fallback>`). Empty values are no-ops. Add | |
| # real per-shape overrides here as needed. | |
| - shape: c4a-standard-72 | |
| parallel_count: '' | |
| zone: '' | |
| zones: '' | |
| steps: | |
| # this isn't a great way to handle this, but GH actions has limitations... | |
| - name: Skip certain jobs | |
| id: checker | |
| if: (matrix.shape == 'c4a-standard-96-metal' && inputs.version_major == '8') | |
| run: | | |
| echo "Skipping ${{ matrix.shape }} for version_major ${{ inputs.version_major}} due to incompatibility." | |
| echo "skip=true" >> $GITHUB_OUTPUT | |
| exit 0 | |
| # we don't need the checked out files, but this is required for the google auth action to work | |
| - uses: 'actions/checkout@v6' | |
| if: steps.checker.outputs.skip != 'true' | |
| - id: 'google-auth-image-testing' | |
| uses: 'google-github-actions/auth@v3' | |
| if: steps.checker.outputs.skip != 'true' | |
| with: | |
| workload_identity_provider: 'projects/527193872801/locations/global/workloadIdentityPools/github-actions/providers/github' | |
| service_account: 'github-actions-image-testing@almalinux-image-testing-469421.iam.gserviceaccount.com' | |
| - name: 'Set up Google Cloud SDK' | |
| if: steps.checker.outputs.skip != 'true' | |
| uses: 'google-github-actions/setup-gcloud@v3' | |
| - name: Download custom CIT image | |
| if: inputs.cit_git_repo != '' && steps.checker.outputs.skip != 'true' | |
| uses: actions/download-artifact@v7 | |
| with: | |
| name: cit-custom-image | |
| - name: Load custom CIT image | |
| if: inputs.cit_git_repo != '' && steps.checker.outputs.skip != 'true' | |
| shell: bash | |
| run: gunzip -c cit-image.tar.gz | docker load | |
| - name: 'Run Google cloud-image-testing tests on ${{ matrix.shape }}' | |
| if: steps.checker.outputs.skip != 'true' | |
| uses: ./.github/actions/cit-run-with-retry | |
| with: | |
| image: "${{ needs.init-data.outputs.image_path || inputs.image_override }}${{ inputs.image_override == '' && '-arm64' || '' }}" | |
| filter: '^(cvm|livemigrate|suspendresume|loadbalancer|guestagent|hostnamevalidation|imageboot|licensevalidation|network|security|hotattach|packagevalidation|ssh|metadata)$' | |
| shape_flag: "-arm64_shape ${{ matrix.shape }}" | |
| parallel_count: ${{ matrix.parallel_count || '1' }} | |
| first_attempt_zone: ${{ matrix.zone || '' }} | |
| zones: ${{ matrix.zones || env.GCP_DEFAULT_ZONES }} | |
| creds_path: ${{ env.GOOGLE_GHA_CREDS_PATH }} | |
| quota_log_file: ${{ runner.temp }}/quota-failures.jsonl | |
| image_ref: ${{ inputs.cit_git_repo != '' && format('cit-custom:run-{0}', github.run_id) || 'gcr.io/compute-image-tools/cloud-image-tests:latest' }} | |
| - name: Upload quota-failure log | |
| if: always() && steps.checker.outputs.skip != 'true' | |
| uses: actions/upload-artifact@v7 | |
| with: | |
| name: quota-failures-pershape-aarch64-${{ matrix.shape }} | |
| path: ${{ runner.temp }}/quota-failures.jsonl | |
| if-no-files-found: ignore | |
| retention-days: 7 | |
| # Aggregates the per-job JSONL quota-failure logs into a single markdown | |
| # table on the run's summary page. Useful for filing quota-increase | |
| # requests: groups by (location, quota, shape) so the highest-impact | |
| # caps surface first. | |
| summarize-quota-failures: | |
| name: Summarize quota failures | |
| needs: | |
| - test-gcp-initialtest | |
| - test-gcp-nonpershape | |
| - test-gcp-pershape-x86_64 | |
| - test-gcp-pershape-aarch64 | |
| if: always() | |
| runs-on: ubuntu-24.04 | |
| steps: | |
| - name: Download all quota-failure artifacts | |
| uses: actions/download-artifact@v4 | |
| with: | |
| pattern: quota-failures-* | |
| path: quota-failures | |
| merge-multiple: true | |
| - name: Write summary | |
| shell: bash | |
| run: | | |
| set -uo pipefail | |
| shopt -s nullglob | |
| files=(quota-failures/*.jsonl quota-failures/**/*.jsonl) | |
| if (( ${#files[@]} == 0 )); then | |
| { | |
| echo "## Quota failure summary" | |
| echo | |
| echo "No quota failures recorded in this run." | |
| } >> "$GITHUB_STEP_SUMMARY" | |
| exit 0 | |
| fi | |
| cat "${files[@]}" > all-quota-failures.jsonl | |
| total=$(wc -l < all-quota-failures.jsonl) | |
| { | |
| echo "## Quota failure summary" | |
| echo | |
| echo "**Total quota failure events:** ${total}" | |
| echo | |
| echo "Group key: (location, quota, shape). Counts include retries." | |
| echo "Use this to scope quota-increase requests." | |
| echo | |
| echo "| Location | Quota | Shape | Count |" | |
| echo "|----------|-------|-------|-------|" | |
| } >> "$GITHUB_STEP_SUMMARY" | |
| jq -s -r ' | |
| group_by([.location, .quota, .shape]) | |
| | map({ | |
| location: .[0].location, | |
| quota: .[0].quota, | |
| shape: .[0].shape, | |
| count: length | |
| }) | |
| | sort_by(-.count, .location, .quota, .shape) | |
| | .[] | |
| | "| \(.location // "?") | \(.quota) | \(.shape // "?") | \(.count) |" | |
| ' all-quota-failures.jsonl >> "$GITHUB_STEP_SUMMARY" |