Image hosting is expensive, so I was wondering what is the financial situation like, and how can we keep this “crucial” (obviously) instance alive.
Image hosting is expensive, so I was wondering what is the financial situation like, and how can we keep this “crucial” (obviously) instance alive.
Removed by mod
There might be a way to point media post assets to an s3 bucket on the backend and have that hosted somewhere with cheap storage like Backblaze, but I definitely doubt that’s been built out yet. Though I admittedly haven’t looked through the code at all.
Removed by mod
Yeah that probably could be done on the server itself, depending on how the media is split up on the filesystem. Say for instance the
/app/mediadirectory is an s3 bucket mounted on the host. Might present some performance problems with doing it that way though.I checked my options and decided to use Cloudflare R2 for this. Lemmy currently supports S3 but is unstable. I’ll wait for it to become mainstream.
Also, I don’t believe it is stable enough to mount S3 to the folder with Rclone.
I haven’t waded through the config to look yet, but glad to see you found something that works for your instance. Lurked your profile a bit (sorry) and saw that pict.rs was able to point to object storage, did you end up doing that already?
Nope. Right now the pict-rs version that supports S3 is not stable yet. I tried to migrate on staging server tho. It worked there. I’ll migrate as soon as pict-rs goes stable.
Nice! Looks like the new version will solve part of the storage issue once it drops. I’ve also been reading reports of “database bloat” but have very few details to go on. Have you been seeing any other storage related problems besides media?
No its mostly on media part