Incident with Pull Requests and Issues – Resolved | Hacker News Reader

Incident with Pull Requests and Issues – Resolved

githubstatus.com

57 pointsarm3216 comments

2-3 9's of availability on most services shown on their status page (and these are always generous). For a platform that wants to be the industry standard CI/CD platform, there is no way you can justify that.

So much software engineering going on

10x more we could say

This is a platform in chronic decline.

After a week ago [0] a week before that [1] and again a few days before that [2] with it is no wonder that each time GitHub falls over, confidence in the platform is rapidly going away.

Sounds like it is once again a great time to self host, as predicted 6 years ago. [3]

[0] https://news.ycombinator.com/item?id=49199446

[1] https://news.ycombinator.com/item?id=49040940

[2] https://news.ycombinator.com/item?id=48973653

[3] https://news.ycombinator.com/item?id=22867803

Decline? AI coding agents drove commits up 14× in a year. Pull requests opened by AI coding agents surged from roughly 4 million in September 2025 to more than 17 million in March 2026 — a 325% increase in six months. [1]

[1] https://www.techtimes.com/articles/318481/20260616/githubs-a...

That's a measure of the costs GitHub incurs to operate, not of its value as a platform.

Yes. Decline.

Even outside of the reliability issues thanks to the AI coding agents, that money is not going to GitHub. It is going to the model creators.

GitHub failed to capture any of that and is even shutting down GitHub Spark. [0] (Lovable competitor) and is also shutting down GitHub Models (Huggingface competitor). [1]

Both products have been a total failure and even if it was available it would have to shut down away to save the platform from further degradation.

[0] https://github.blog/changelog/2026-08-04-upcoming-deprecatio...

[1] https://github.blog/changelog/2026-07-30-github-models-is-no...

Why are you acting like 14x is 1000x?

325% is just 3.25x. None of these numbers are impossible to scale to.

Do you think these outages are happening because servers are just all too busy?

Just scale it up, you have Azure. Come on. The problem isn't the scale, it is the poor software engineering.

“You need to use cloud services because the uptime can’t be matched if you run it yourself”

I mean it’s true, but not in a good way.

Also i mean... Putting your software forge with cicd on publicly accessible infra isn't the wisest security posture.

No one else is going to basically give everyone free git hosting with such generosity as Microsoft.

I think LMMs have made it impossible for unlimited free repos for everyone to work.

Generosity lol. They get a ton out of GitHub, not the least being all that sweet sweet code to train their AIs on.

If you think Microsoft is generous, you are extremely naive. LLMs and their produced content are less interesting to MS because they don't present original human data for training.

The incident is already superseded.. Or maybe they fixed it by shutting down all web gateways..

It's utterly unreliable. I laughed when I read the title. We moved our company to self hosted Forgejo a while ago and literally the day after, there was a major disruption on GitHub. Their service also isn't that great even when it works. We decided to move because they abrupt ly disabled all our LFS access without any prior warning causing a private outage for our developers. Fun!

PSA: You can still create PRs with the gh cli tool.

Reply on news.ycombinator.com