The bridge to agent-scale Git infrastructure

Thanks I just want my GitHub actions to not break every so often though.

• da_chicken · 4 hours ago

  I thought breaking every so often was already the dominant feature of GitHub actions.

  • intothemild · 3 hours ago

    Yeah how else am I supposed to take a break or an afternoon nap

  • onion2k · 2 hours ago

    Chaos engineering as a service.

    • skrtskrt · 7 minutes ago

      Whereas functioning CI is just arbitrary shell scripts with glue and tape as a service.

  • skrtskrt · 10 minutes ago

    When it's broken is the only time it might not be an active security threat

• Kinrany · 1 hour ago

  I'll settle for git operations and PR merges not breaking.

This is so tone deaf in timing, it could only be microsoft, and it (weirdly) makes me feel better.

• lemonish97 · 3 hours ago

  Why is this "tone deaf in timing"? Isn't this addressing gh's biggest bottleneck at the moment?

  • throwaway69123 · 12 minutes ago

    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.

    • skrtskrt · 8 minutes ago

      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

• philipallstar · 2 hours ago

  Commentary on PR strategy is really low quality commentary.

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...

• laura2013 · 16 minutes ago

  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

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.