On April 27, 2026, Github experienced a major outage that lasted for several hours. The outage affected the merge queue, which is a major part of the Github workflow. According to the CTO,
In some case, squash or rebase commits were generated from the wrong base state, making earlier changes appear reverted in branch history.
I honestly want to know how this could happen. The merge queue picked a wrong base? Wouldn’t git have errored? Was github doing --force under the hood?
Trunk.io has an image that makes things clearer:
What was supposed to happen, and indeed, what the UI suggested did happen is that the commits on the branch that was to be merged were squashed, then merged to the tip of the main branch, perserving the correct history.
Instead, the temporary branch selected the point at which the branch was forked and merged that (plus your changes) to the main branch,losing everything else up to the point of merging. Nice!
Oh, and the UI lied to you about what was going on. To you, everything appeared normal. It was only when you started shipping features again when you noticed. RIP your productivity that day.
It takes genuine talent to screw up this bad.
Github went from the central place where development and collaboration happens to a place where your changes could be reverted, the service could at any time just not work, and there are random security vulnerabilities that apparently leaks your code.
You know it’s bad when you have to refer to downtime incidents with dates because they occur so often in a year.