Retry Failed Matrix Jobs #163
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: Retry Failed Matrix Jobs | |
| permissions: | |
| actions: write | |
| contents: read | |
| env: | |
| FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: true | |
| on: | |
| workflow_run: | |
| workflows: [ "Test", "Sanitize", "Deploy", "Binary Tarballs", "Release Tarballs" ] | |
| types: [ completed ] | |
| jobs: | |
| rerun-failed: | |
| name: Retry failed matrix jobs | |
| if: >- | |
| ${{ | |
| github.event.workflow_run.head_repository.full_name == github.repository && | |
| github.event.workflow_run.run_attempt == 1 && | |
| ( | |
| github.event.workflow_run.conclusion == 'failure' || | |
| github.event.workflow_run.conclusion == 'timed_out' | |
| ) | |
| }} | |
| runs-on: ubuntu-latest | |
| steps: | |
| - name: Re-run failed jobs when most matrix jobs passed | |
| uses: actions/github-script@v8 | |
| with: | |
| script: | | |
| const run = context.payload.workflow_run; | |
| const owner = context.repo.owner; | |
| const repo = context.repo.repo; | |
| const workflowName = run.name; | |
| const excludedByWorkflow = { | |
| "Test": [ | |
| "Create job matrix", | |
| "Test - summary", | |
| ], | |
| "Sanitize": [ | |
| "Create job matrix", | |
| "Sanitize - summary", | |
| ], | |
| "Deploy": [ | |
| "Workflow Overview", | |
| "Create Job Matrices", | |
| "Mirror images to Docker Hub", | |
| "Update Docker Hub description", | |
| "Deploy - summary", | |
| ], | |
| "Binary Tarballs": [ | |
| "Attach GEMC tarballs to dev release", | |
| "Create Job Matrix", | |
| ], | |
| }; | |
| const excluded = new Set( | |
| (excludedByWorkflow[workflowName] || []).map((name) => name.toLowerCase()), | |
| ); | |
| const jobs = await github.paginate(github.rest.actions.listJobsForWorkflowRun, { | |
| owner, | |
| repo, | |
| run_id: run.id, | |
| per_page: 100, | |
| }); | |
| const countedConclusions = new Set([ | |
| "success", | |
| "failure", | |
| "cancelled", | |
| "timed_out", | |
| "action_required", | |
| ]); | |
| const matrixJobs = jobs.filter((job) => { | |
| const name = job.name.trim(); | |
| return job.status === "completed" | |
| && countedConclusions.has(job.conclusion) | |
| && !excluded.has(name.toLowerCase()); | |
| }); | |
| const successCount = matrixJobs.filter((job) => job.conclusion === "success").length; | |
| const failedCount = matrixJobs.length - successCount; | |
| const successRatio = matrixJobs.length === 0 ? 0 : successCount / matrixJobs.length; | |
| core.info(`Workflow: ${workflowName}`); | |
| core.info(`Run: ${run.html_url}`); | |
| core.info(`Matrix jobs: ${matrixJobs.length}`); | |
| core.info(`Matrix successes: ${successCount}`); | |
| core.info(`Matrix failed/cancelled/timed out: ${failedCount}`); | |
| if (matrixJobs.length === 0) { | |
| core.notice("No matrix jobs were found for this workflow; not retrying."); | |
| return; | |
| } | |
| if (failedCount === 0) { | |
| core.notice("No failed matrix jobs were found; not retrying."); | |
| return; | |
| } | |
| if (successRatio < 0.5) { | |
| core.notice("Fewer than 50% of matrix jobs succeeded; not retrying failed jobs."); | |
| return; | |
| } | |
| await github.rest.actions.reRunWorkflowFailedJobs({ | |
| owner, | |
| repo, | |
| run_id: run.id, | |
| }); | |
| core.notice("Requested a re-run of failed jobs for this workflow run."); |