Another thing that I thought of is that maybe I can somewhat replicate NixOS’ rollback feature, which is my absolute favourite feature of it, using a combination of Git and ZFS snapshots? I’d like to hear your thoughts on this.

While I don’t know what exactly NixOS’ rollback feature entails, I run / on btrfs and use a cronjob to create snapshots of the “rootfs” subvolume. I maintain a history for the last 48h. This allows me to roll back when something goes wrong, which basically never happens though. That way though there is basically nothing to fear when doing large updates.

Check this out: https://lemmyverse.net/communities?order=subscribers

Not an expert on Lemmy, but:

So if a instance goes down, does that mean all accounts would be gone?

Essentially, yes, all accounts of that instance would be gone.

Would people lose all post/comments regardless in which community/instance they posted?

No, posts/comments on other instances would still be available.

Unfortunately, my previous instance has had some issues with stability and database corruption repeatedly. Lost posts, subscriptions etc. I am here now due to the similar federation policy, which I believe is the right one. The performance is also great. Very promising!

This thread is an interesting “filter bubble” experience. Here, a city in central europe, nobody I know wears one anymore, even those who were always extra-careful on the cautious side. Basically nobody in stores wears one nor in public transport. Yeah, on occasion you find 1-2 exceptions that confirm the rule.

Probably, this thread is largely visited by those who still do, and ignored by those who don’t.