some of the performance problems we're seeing on cybre.space could be due to the aborted 3.0 migration that we tried yesterday -- this morning we're seeing a lot of CPU usage from postgres but not a lot of disk usage. we're going to try to complete the v3.0 upgrade to rule out the possibility that the database is in an unclean state. there may be some downtime as this progresses.
thanks for being patient as we troubleshoot things!
cybre.space is having service problems right now "thanks" to our hosting provider, but i want to reassure everyone here that we have no intention of taking the instance down entirely, and we have appropriate backups just in case the worst happens (which have been tested to be restorable). cybre.space isn't going anywhere.
hmm still seeing occasional 504s here
i feel like SSD backed storage should be able to handle more I/O than i was seeing, though. interested to see where my support ticket goes.
if things really have become stable again it might just be that postgres was hitting the disk too hard. if that's the case then it might be prudent to take the server down to do the migration for 3.0
i think i'm going to leave it for tonight and retry the 3.0 upgrade tomorrow under the circumstances
the instance seems stable for some reason but i'm scared to touch it
judging from the steadily increasing error rate in netdata my suspicions seem to be accurate
things are sort of working now but i suspect it's not actually fixed
the disk on our hetzner vserver seems to have severely degraded performance, so i tried upgrading to the next vserver size to see if we'd get a different physical disk or something. i dunno.
i don't know what happened but the instance seems to be working again *knocks on wood*
admin of https://cybre.space.
dragon whomst do the software
Cybrespace is an instance of Mastodon, a social network based on open web protocols and free, open-source software. It is decentralized like e-mail.