I’m running a small instance just for myself and a friend. We basically use it to access the Lemmy fediverse.
I was under the impression that our instance would not download imagery that was not hosted on it but the pict-rs “files” directory ended up with 45 gig of data which ultimately filled up the VPS storage space.
Is there a command that I can automate to prune this regularly or does anyone have a solution?
Sadly not yet. Hopefully this will be improved in the future.
However, this sounds like you used one of these follow everything bots, which is just a bad idea as it results in these kind of issues.
I did but “all” is useless without it and having to use a separate website to discover communities just doesn’t make sense.
I think at some point I will figure something out.
I just took a look at the repository for pict-rs and unfortunately nothing in the official docs about cleaning up images. Did you happen to use a bot to assist you in finding communities? If so this might be the reason. After two weeks of running, my image size is only 2.2mb. I should also ask you what the size of your database is. The activity table can get awfully big and very quickly.
I did use a bot…
Ah, I see. I hope some maintenance tools will be forthcoming.
Imagine if media in Lemmy was all hosted in a distributed network filesystem like Iroh, where instances only function as inserters and exit nodes for that media.
This way, smaller instances can have a smaller cache corresponding to the media that was actually needed by it (recently). And independent peers can help by participating in the distributed file-system network without running instances themselves.
Woah, that seems really neat! What would be the cons to doing this (other than the implementation time and effort of course)?
Can you just run a cronjob to delete files in that directory every day?
(Maybe there’s a reason you can’t do this, I don’t know how Lemmy instance works)
It would definitely be nice to have some image retention policy. If an image hasnt been accessed in say 6 months its probably safe to say I won’t care about it. And if need be it can pull it from the original server again.
