One of the many reasons I moved to self hosting. I use ZFS to backup every 15 minutes, ... could do it even more frequently but that seems a little pointless.
Also moved away from Gitlab because it's so damn slow.
It for sure has an excruciatingly slow UI, but I haven't yet been able to stomach the Ruby magick enough to dig in and find out if it's all of GitLab that's slow, and thus their culture of who-cares just propagates to the UI, or it's literally just the web frontend that's molasses yet because so many folks interact with it over the web that's where the conclusion ends up
I know their backend git proxy is written in golang, their runner agent is written in golang, it spawns CI jobs using containerd, written in golang, and they use postgresql and a redis-esque KV -- although for that part I do believe they're still using Sidekick (ruby) for doing job dispatch, so that one could very easily lolol back into not actioning tasks efficiently
GitHub Actions sure does enjoy just sitting there twiddling its thumbs when I push "run job," and is also both Ruby and their own crazypants ajax-y web framework, so I don't think it's exactly the shining star of performance itself
Gitea, i can't recommend it enough. Lightening fast compared to the proprietary offerings, clean simple UI like an older Github, and super simple to self host.
So long as you have an existing CI/CD story, or are willing to invest in head-to-heading them, to find one that fits your needs. Because both it and Forgejo's "we're deploying act, what could go wrong" give me the heebie-jeebies. And, aside from that, there are so many FLOSS ones they could have chosen that legitimately work making that decision extra opaque to me
This was my thought too, but I figured I just didn't understand the problem. Why use git commands to backup when a file system copy seems like it would do? zfs snapshot takes less than a second on any size repo. zfs send transfers just the changes since the last backup, as fast as your network, more or less.
Yup, once you've used snapshots for backups once, doing any kind of filesystem level backup just seems absurd. I now use it as the basis for every piece of infrastructure I add that holds valuable data. The only place it doesn't really fit is distributed databases.
I know their backend git proxy is written in golang, their runner agent is written in golang, it spawns CI jobs using containerd, written in golang, and they use postgresql and a redis-esque KV -- although for that part I do believe they're still using Sidekick (ruby) for doing job dispatch, so that one could very easily lolol back into not actioning tasks efficiently
GitHub Actions sure does enjoy just sitting there twiddling its thumbs when I push "run job," and is also both Ruby and their own crazypants ajax-y web framework, so I don't think it's exactly the shining star of performance itself