Gregory's ServerMy #GoToSocial instance is back up. Here's a helpful reminder to fellow admins: don't be like me and leave Monit watching over your site. The migration takes longer than the 10–15min that my Monit config would allow, restarting the server even though migration wasn't finished.
https://alexschroeder.ch/view/2024-10-27-upgrade-gotosocial
 5 comments
Another strange effect of my #GoToSocial upgrade, perhaps coupled with my other problems – and I really don’t know who is to blame: the Toot! app I use just showed me all my recent direct conversations as “new” (with those weird bubbles along the left side). I don’t think the Mastodon Client protocol offers a “seen” flag so perhaps the messages all got new ids? I have no idea. It’s not very annoying just surprising. I kept updating my blog post at https://alexschroeder.ch/view/2024-10-27-upgrade-gotosocial … current status is that the system runs, there are no more sqlite errors, as far as I can tell, but there are still occasional lock-ups that I have no explanation for. Monit restarts the instance if the website is down for half an hour (6 failed checks in a row with 5 minutes between checks). I've increased the memory allocation by systemd to high 600M and max 1000M; systemd is reporting about 315M being used, currently. |
Another thing to note for the #GoToSocial upgrade is that I ran 16.0 using a ssystemd MemoryMax of 200M; today the upgraded instance with 17.1 ran fine for a while and then locked up. A restart didn't bring it back. It remained stuck after a log message saying something about compiling WebAssembly. I increased MemoryMax to 300M, no change. I increased it to 500M and the instance came up. Just in case you're as memory-stingy as I am...