I have a process which copies emojos from the hajkey blahaj.zone site to our lemmy.
Last night (for me) it copied an invalid emojo into the database and it broke lemmy for hours while Ada and I were sleeping.
I've found the issue and fixed it now, so everything should be purring along once again, but for those who were left without the instance to use, I sincerely apologise.
No worries, things happen — and you’re community-building volunteers, not an on-call rotation. 💜 I’m just glad we’re back now! Thanks for your efforts during these past 24 hours. 🏳️⚧️😊
It wasn't an actual emojo. The script processed the SQL header column names as an emojo and tried to add them. Unfortunately publicUrl is not a valid URL, so lemmy's /api/v3/site metadata endpoint started returning an error relative URL without a base instead of the JSON that the website was expecting and so the site just stopped working for everyone the next time it tried to load that url.
So glad it‘s up again. I was thinking it was something like that since it kept giving a parsing error when I was trying to log in. Good that it‘s fixed now.
{"data":{"msg":"Couln't upload file, Error in DB, IO error: Os { code: 28, kind: StorageFull, message: "No space left on device" }","files":null},"state":"success"}
Hmm, still getting errors, for example when trying to upload profile images