Thread with 15 posts
jump to expanded post
EXCITING “hikari runs her own instance” UPDATE!!!
my server ran out of disk space again and THIS time it took the database with it! i just had to restore the state of my server to what it was four days ago (last backup). four days of posts… gone. luckily none of them were good posts 😌
but now we're back. we're so back. or whatever. total downtime less than an hour probably? :p
i posted about it on the formerly birded side: https://twitter.com/hikari_no_yume/status/1747708548774076781
hazards of running your own instance include not only media caches filling up your disk but also the log file filling up your disk!!
@hikari simply do not keep logs, what could go wrong
@hikari oh! oops. well, welcome back.
@irenes thank you!
@hikari Set up some monitoring and daily backups?
@hikari Thinking about git and distributed restoration. In theory I have all your public and unlisted posts, at least the non-replies! If only I could send them to you.
@jrose something like secure scuttlebutt could definitely gracefully recover in a situation like this
@hikari Right, plain old ActivityPub doesn’t have any auth, so you’d have to trust that I’m not going to pretend you said “WIN AN IPAD FROM HIKARI REPLY HERE”
@jrose ActivityPub is obsessed with signatures but i guess the posts themselves aren't signed?
wait how the hell do reposts work then
@hikari It's interesting that I can still see some of the posts that you lost, I guess cause my instance caches them or something, but the direct link to the post gives a 404
@quirk the fediverse is full of ghosts
@hikari I'm imagining a future fediverse (or other decentralised social media) where posts are signed by the posters private key, so the instance can ask other instances for that they have in their cache to restore after an outage, knowing that the posts have not been tampered with.
@saagar we must solve “catch up on the last few days of posts” alignment