I think this is perhaps somehow related to Volumes because I was just able to recreate it.
I looked at the File Manager for the Volume and spotted that some folders and the metadata.db were owned by root instead of Cloudron.
I started changing them one by one via the File Manager but that was tiresome so I open the Web Terminal and ran:
chown -R cloudron:cloudron /library
That fixed the issue on another test I installed, but didn't resolve the issue with my initial install. So I just kept the next test install and moved it books.uniteddiversity.coop which is now working again 🙂 (well, working in a unstable and outdated state 😉 )
I'm pretty sure the issue was created when I rsync'd up a load more books to the Hetzner Storage Box where the Volume actually lives. I'd forgotten that I need to do the chown thing afterwards (and now I'm wondering if I can edit my rsync command to stop me having to chown afterwards...)
@jdaviescoates Yes, I think that would be the next step. An alternate angle is to bundle ice server as part of ampache app package itself and make it optional. But I don't have experience with hosting ice cast channels to know what is the better approach. Does anyone else here know?
/me goes to reduce the size of the volume back to a perfectly sufficient 10 GB...
Oh, bollocks. You can't decrease the size of Hetzner Cloud Volumes, only increase them. So I'd have to create a whole new one and then and then and then...can't be bothered with all that so I guess I'll just stomach the slightly increased cost of 15 GB over 10 GB.
Honestly, having been in ecommerce development for 20 years, I'd say most small sites would be better off using https://ko-fi.com/.
If you really want self-hosted, WP & Woo is my preferred and main experience - but it was still very dev needy.
I don't know why Odoo hasn't made it to be packaged yet but I'd be more inclined to look at that than any of the other options.
Like I say, 20 years experience on the ecommerce merry-go-round, to the point that I barely bother explaining my recommendations now, so it's just a bit of experience from an old man as to how much of a time-hole the niche options can become.
+1 for this. The fact that the docs warn strongly against installing Nextcloud apps makes Nextcloud's whiteboard app a no-go for me (unless that's not the case with this app in particular). My team and I are looking for something more robust (media support, etc.) than the currently-offered WBO app. We currently use Miro.
@nebulon I thought it was perhaps the Broken Links Checker on one of my WordPress installs, as as @marcusquinn noted it's widely known to sometimes cause such issues, but I've disabled that and I'm still getting these /mysql was restarted (OOM) (although they don't seem to appear in the event log?).
I note that the most recent one happened 5 minutes after my backups are due to start, and the other previous two times it's been 7-9mins before updates are due to run.
I wonder if either of those (backup/ update) processes might have something to do with it?
I guess I could just give mysql more memory and not worry, but be nice to know what's happening and why...