When the merge queue misbehaves: what to check and how to recover

A merge queue tests each pull request together with the ones ahead of it and merges them in order. It is a strong feature, and like any shared system it has had bad days. Knowing the common failure modes lets you check for them.

An incident worth knowing about

On April 23, 2026, according to GitHub's availability update, pull requests merged through the merge queue with the squash method could produce wrong merge commits when a group contained more than one pull request. Changes from earlier merges were reverted by later ones. GitHub reported 658 repositories and 2,092 pull requests affected, and said the merge and rebase methods were not affected.

Flaky required checks

A required check that fails at random removes pull requests from the queue and delays everything behind them. Teams commonly turn on the queue, hit a flaky test within days, and turn it off. Separate the checks that must pass from the ones that are only informational, and quarantine tests that are known to be flaky.

Timeouts and stuck entries

Community reports describe queue entries that time out and are removed while the underlying job carries on, and queues that stall for long periods. If a pull request sits in the queue much longer than your checks take, look at the queue's status and the checks before waiting longer.

A short checklist after queued merges

Queue and scheduler do different jobs

A merge queue decides order and tests combinations. It does not decide the time. A scheduled merge decides the time. You can use both, and a deploy watcher with an automatic revert is a good safety net behind either.

Do this in Merge Scheduler: Schedule your first merge

Related: GitHub auto-merge, merge queue and scheduled merge: what each is for · Merge commit vs squash vs rebase: what each GitHub merge actually does · How to automatically revert a failed deploy (and when not to)