Report #75871
[bug\_fix] Matrix job failures cause immediate cancellation of all other running matrix jobs
Add \`strategy: fail-fast: false\` to the job definition to allow all matrix combinations to run to completion regardless of individual failures.
Journey Context:
You configure a workflow with a matrix strategy testing your application across Python 3.8, 3.9, and 3.10 on Ubuntu, Windows, and macOS—a 9-job matrix. You push a commit that introduces a Windows-specific bug. The Windows \+ Python 3.8 job fails within 2 minutes, but you notice that the macOS \+ Python 3.10 job \(which was running for 10 minutes\) is immediately cancelled and marked as skipped with the annotation "The job was canceled because another job in the matrix failed." You wanted to see if the bug was Windows-specific or affected other platforms, but now you have no results for the other 8 combinations. You check the Actions UI and see all jobs are greyed out except the one that failed. You look at the workflow YAML and see \`strategy: matrix: ...\` but no \`fail-fast\` setting. You search the GitHub documentation and discover that \`jobs..strategy.fail-fast\` defaults to \`true\`. When true, GitHub Actions automatically cancels all in-progress and queued jobs in the matrix as soon as any single matrix job fails. The debugging involves realizing that the cancellation is not due to resource limits but is an intentional default behavior. The fix is to explicitly add \`fail-fast: false\` to the strategy block. This allows each matrix combination to run to completion independently, providing you with the full test matrix results and showing exactly which specific combinations fail without cancelling the others. This is essential for identifying platform-specific or version-specific failures.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-21T09:56:41.639794+00:00— report_created — created