Delivery & CI · Updated 2026-09-17
Merge queue
A system that serialises merges, testing each change against the current tip of the main branch before it lands, so green pull requests cannot break the branch on arrival.
Also called: Merge train
What it is
Two pull requests can each pass CI against an older main branch and still break it when both land. A merge queue removes that class of failure by rebasing and testing changes in order, merging only what actually passes against the real target.
Why it shows up in code review tooling
Merge queues and review automation compete for the same moment in the workflow — the gap between approval and merge — and they interact. If a review tool posts a required status check, that check has to pass in the queue as well as on the branch, which means a slow or flaky reviewer becomes a throughput problem for the whole team.
Why it matters when you are evaluating
If you already run a queue, ask how the tool’s checks behave inside it: does the review re-run on the rebased commit, does it re-bill you for that run, and what happens to a finding the author already resolved? Tools designed around a single PR event sometimes behave oddly here.
Common mistakes
- Adding a blocking AI check to a merge queue before knowing its p90 latency.
- Paying twice for the same review because the queue triggers a fresh run per rebase.
- Using a queue to compensate for a slow test suite rather than fixing the suite.