GitHub Status - Incident with Pull Requests

[-]

Sigmatics@reddit

No day without an incident these days

[-]

RGBrewskies@reddit

I want to believe it's not because they're letting AI program the fucking thing, but we all know it's because they're letting AI program the thing.

[-]

Ancillas@reddit

It’s probably because AI has amped up the amount of code being pushed, so they’re having to allocate resources to fixing infrastructure and scale up issues.

[-]

thatsnot_kawaii_bro@reddit

That, but surely nothing to do with the AI quotas they push on devs

[-]

yawaramin@reddit

It’s probably because AI has amped up the amount of code being pushed

Yet the product has not noticeably improved.

[-]

Kuinox@reddit

I think the comment you reply to, speak about the amount of commit pushed by customers, not by their engineers.
Commit count have been exploding on github theses past months.

[-]

dangerbird2@reddit

It’s because they’re letting Microsoft program the thing

[-]

AlyoshaV@reddit

This incident was 10x worse than just downtime, though

[-]

OrkWithNoTeef@reddit

This regression given to you by a regression machine

[-]

cosmic-parsley@reddit

We have resolved a regression present when using merge queue with either squash merges or rebases. If you use merge queue in this configuration, some pull requests may have been merged incorrectly between 2026-04-23 16:05-20:43 UTC.

Anyone know what the incorrect merges actually looked like?

[-]

Ours reverted prior commits while merging code. Meaning the codebase ended up in a state that we had never intended and had never gone through CI. Fortunately everything it reverted ended up being inconsequential but I'm shocked that this story hasn't been picked up.

[-]

732@reddit

Ours just didn't merge the code

[-]

ZirePhiinix@reddit

Because automated checks did not validate merge correctness for multi-PR squash groups, the defect surfaced only in production.

What a terribly written statement. I don't have anything to do with Github but I can write a better version.

Because automated checks didn't cover this scenario, it wasn't caught during test and broken code got deployed to production. Subsequent testing showed that proper tests would've caught the problem.

[-]