The Missing GitHub Status Page
119 points by skade
119 points by skade
93.11℅ uptime, Isn't that... kinda low?
GitHub stopped updating its status page with aggregate uptime numbers some time ago — if you use it regularly, you might have a feeling why. This is the missing version.
I guess that's what's implied on this page.
The two services with individual single-9 uptime are Actions and Copilot; the rest of the services individually do have two-9s uptime in the last 90 days. Copilot is the worst at 96.88%, currently.
Actions is just bad, we know that. Copilot I don't really care about. So on aggregate, depending on you interests, that 93 (currently 92)% uptime isn't quite as bad as it looks.
(Also, how did that ℅ "care of" symbol end up in your comment?)
I remember people saying that Codeberg is quite unstable. Looking at their status page is inconclusive as it gives only uptime for the last 14 days, which is 98.43%. For GitHub we have it for 90 days so it can't be compared directly. Does anyone have some better figures for Codeberg?
I mean, it's aggregated across all services where any tiny service degradation considers the whole platform broken. However, i think that's how they reported their uptime in the original variant of this page (I'm still trying to find a snapshot of it).
So your personal uptime experience might be different.
The degraded uptime is probably a side-effect of them aggressively migrating their infrastructure to Azure.
https://thenewstack.io/github-will-prioritize-migrating-to-azure-over-feature-development/
Years ago (as in, somewhere in 2017/2018) GitLab.com moved away from Azure precisely because of how garbage it was. In fact, the only reason we were (for a short time at least) using it was because of the free credits we got through Y Combinator. I don't remember a single developer or SRE having something good to say about our time using Azure.
That was before Microsoft bought GitHub. When it happened, the Azure folks were told that they needed to migrate GitHub to Azure, but this wouldn't happen until the GitHub people signed off on it, so Azure needed to improve their services to meet GitHub's requirements.
It's starting to be a huge pain in the ass at work: I had the feeling like we faced at least one outage a week and these graphs confirms it...
It such a recurring pain that we are starting to talk about moving to self hosted solutions like Forgejo. I guess it's the benefit of it :')
Isn’t it a problem that this page uses GitHub’s infrastructure to be shown? 🤔
Well apparently not.
That just means the site can be down sometimes and maybe the Actions can’t pull from GitHubStatus but since it’s based on past data, shouldn’t be a problem and data should still be accurate.
I'm aware that the correct thing is to open a PR to add the link to https://github.com/mrshu/github-statuses/blob/master/site/index.html#L116 but in lieu of that 👈 is the repo
I think it may have been implied by their mention of whatever "Flat Data" is supposed to do, but eventually no sane person would be able to clone the repo since it writes back an .atom snapshot followed by several csv, json, and jsonl files to itself as if .git is a generic blob store https://github.com/mrshu/github-statuses/commits/master
to make the world come full circle, I'd laugh and laugh if eventually it knocked over github actions as it tries to use: github/checkout@whatetever followed by use: github/push@goodluck
eventually no sane person would be able to clone the repo since it writes back an .atom snapshot followed by several csv, json, and jsonl files to itself as if .git is a generic blob store
It will easily be years until that point. Commits by themselves are very lightweight, and differences in text content pack really well in git.