Merge Queue Simulator

GitHub's merge queue runs pull requests with concurrent builds and tests them together before merging into main. This saves time by running CI in parallel, but when a commit fails, the entire active window restarts.

This interactive simulator lets you experiment with different configurations and see how merge queue behavior changes with varying CI success rates, build concurrency, and durations.

The Three Lanes

Queue

Commits waiting to be tested. The top N form the "active window" and run CI in parallel. The rest wait their turn.

Rejected

Commits whose CI failed. When the head of the queue fails, it's ejected and the remaining active window restarts CI. That's the costly part.

Merged

Commits that passed CI and were merged into main. Consecutive successes at the head are merged together in one step.

How the Queue Works

1

The first N commits (build concurrency) run CI simultaneously with random durations.

2

When CI finishes, the queue evaluates from the head (oldest) down.

3

Consecutive successes from the head are merged together.

4

If the head fails, it's rejected and all remaining active commits restart CI, wasting their progress.

This is why a low success rate with high concurrency is so expensive. One failure can waste the CI time of many commits.

Controls & Step Mode

Use the sidebar on the left to configure success rate, build concurrency, CI duration, and speed. Try the presets for quick scenarios.

⏸

Step Mode is on by default. The simulation pauses before each merge or reject, showing you exactly what will happen and its CI time impact. Click Continue (or press Space) to proceed.

Space Start / Pause / Continue
R Reset

Build Concurrency Analysis

The sidebar includes an interactive Build Concurrency Analysis chart that helps you find the sweet spot for your configuration.

C

Cost (red) shows how much more CI you pay as concurrency grows. 1x means running one commit at a time. As success rate drops, higher concurrency wastes more CI.

W

Wall Clock (purple) shows estimated total time to process all commits. Higher concurrency reduces wall clock time through parallelism.

Click or drag on the chart to set your build concurrency. The Wall Clock and Cost values below it update in real time based on your settings.

Cost Tracking & Summary

The stats bar at the top tracks useful vs. wasted CI time, wall clock, waste percentage, and time saved compared to running commits sequentially.

$

The Summary page (top bar button) shows full metrics after the simulation ends, including a cost calculator where you set your CI rate and runner count to estimate real-world costs.

↻

After resetting and running again, the summary shows a comparison of your previous run vs. the current one, so you can see the impact of each change.

Use the Instant button to skip animations and get results immediately. The Glossary in the sidebar footer explains all key terms.

Try a Preset

Click a preset to load it and close this guide. Or click Next to finish without changing settings.

MQ SIM

Simulation Complete

Glossary

GitHub Merge Queue

Queue & Ordering

CI & Execution

Metrics

Scenarios