Skepra / Web recovery & resilience

GitHub Actions workflow starts delayed: protect a frontend delivery queue

A push can be complete while its frontend build has not begun, so treat queued, failed and duplicate workflow states as separate decisions.

Discuss the next recovery or maintenance decision

Scenario long-tail: GitHub Actions workflow does not start during frontend delivery - how to verify queue state and artifact handoff.

On November 16, 2026, GitHub reported degraded Actions performance alongside elevated pull request merge timeouts. The affected workflow-start behavior matters to frontend teams because a push can be complete while the build, visual review or deployment-preparation job has not begun. The public incident record does not prove that any specific workflow or client was affected.

Distinguish queued from failed

For a delayed workflow, record the repository, workflow name, run number, commit SHA, event type, queue timestamp and last visible status. A missing runner log is not the same as a failed test. A job that has not started should not trigger an emergency dependency update or a second release commit.

Keep the client-facing queue explicit. The delivery owner should know whether the next action is to wait, retry, use an approved runner or hand the task to an operator. Avoid launching duplicate workflows until the original run state is known.

Use a recovery worksheet

Use one row for the intended commit, one for the workflow run, one for the expected artifact and one for the client decision. When service recovery is confirmed, retry the original commit once and compare the run number, start time, check conclusion and produced artifact.

If the workflow starts but the artifact differs, the incident is no longer a sufficient explanation and the build itself needs review. Preserve the last known-good artifact and communicate uncertainty as a provider-state issue under review.

Applicability and limits

This applies to repositories using GitHub Actions during the status incident or a comparable workflow-start degradation. It does not establish a GitHub outage outside the recorded component, a broken client build or permission to bypass branch and deployment controls. No client workflow or artifact was observed for this article.

GitHub Status incident history