Report #76538
[bug\_fix] Matrix jobs are cancelled immediately when a single job fails, preventing complete test feedback
Set \`strategy: fail-fast: false\` in the job definition containing the matrix. The fix works because GitHub Actions defaults to \`fail-fast: true\` for strategy matrices, which automatically cancels all in-progress and pending jobs in the matrix as soon as any single job fails, in order to conserve compute resources.
Journey Context:
A QA engineer configures a comprehensive test matrix to validate a library across Node.js versions 16, 18, and 20, and operating systems ubuntu-latest, windows-latest, and macos-latest. They push a commit introducing a platform-specific bug that affects only Windows. The workflow starts, and the Ubuntu-16 job completes successfully. The Windows-16 job starts, encounters the bug, and fails. Immediately, the GitHub UI shows all other jobs \(Ubuntu-18, Ubuntu-20, MacOS-16, etc.\) transitioning to a 'Cancelled' state with a grey icon. The logs for these jobs show 'Canceling since a failure occurred in the matrix'. The engineer needs to know if the bug affects only Windows or specific Node versions, but the cancellation denied that data. Investigating the workflow syntax, they discover the \`strategy\` block defaults to \`fail-fast: true\`. By explicitly setting \`fail-fast: false\`, the next run allows Windows-16 to fail independently while Ubuntu-18, Ubuntu-20, and MacOS variants continue to completion, providing the full matrix of results necessary to isolate the platform-specific issue.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-21T11:03:55.103915+00:00— report_created — created