I'm surprised at the level of detail presented in this post. Say what you will about the product, but I appreciate lifting the veil on what goes on behind the scenes.
These changes seem to be correlated with huge instability. Is the instability caused by the increased use? Seems unlikely given the nature of some of the issues we have experienced. Or is it related to this move to azure infra under the guise of capacity but in reality, a longer term eating of github into the Microsoft mothership.
I'm so tired of them whining about how hard it is to scale due to agents. First of all, GitHub uptime has always been bad, even before the MS acquisition and way before agents. Secondly, plenty of other services have much more demanding scale problems with much better uptime. How many millions of hours of video is YouTube streaming and transcoding? When was the last time YouTube went down? I'm done with the excuses and have moved to self-hosting.
I'm surprised at the level of detail presented in this post. Say what you will about the product, but I appreciate lifting the veil on what goes on behind the scenes.
I wonder if this post was planned prior to Anthropic's announcement regarding Github's inability to handle their volume of merges.
https://www.anthropic.com/research/redesigning-source-contro...
i'd say yes. it's a 3 part series [oct 6](https://github.blog/engineering/architecture-optimization/bu...), next at their conf next week
Thanks I just want my GitHub actions to not break every so often though.
I'll settle for git operations and PR merges not breaking.
I thought breaking every so often was already the dominant feature of GitHub actions.
When it's broken is the only time it might not be an active security threat
Chaos engineering as a service.
Whereas functioning CI is just arbitrary shell scripts with glue and tape as a service.
Yeah how else am I supposed to take a break or an afternoon nap
This is so tone deaf in timing, it could only be microsoft, and it (weirdly) makes me feel better.
Why is this "tone deaf in timing"? Isn't this addressing gh's biggest bottleneck at the moment?
These changes seem to be correlated with huge instability. Is the instability caused by the increased use? Seems unlikely given the nature of some of the issues we have experienced. Or is it related to this move to azure infra under the guise of capacity but in reality, a longer term eating of github into the Microsoft mothership.
It's been a lot worse recently but the increase in instability is fairly well correlated with how long it's been owned by Microsoft.
People have been on here complaining about GitHub being unstable for years and years
Commentary on PR strategy is really low quality commentary.
[flagged]
[flagged]
I'm so tired of them whining about how hard it is to scale due to agents. First of all, GitHub uptime has always been bad, even before the MS acquisition and way before agents. Secondly, plenty of other services have much more demanding scale problems with much better uptime. How many millions of hours of video is YouTube streaming and transcoding? When was the last time YouTube went down? I'm done with the excuses and have moved to self-hosting.