@spiral @wowaname @pengu Sorry for the delay. The main domain terminates to haproxy as a frontend, with several URL rules routing to different backend services (Synapse, GUN). You are indeed probably seeing errors those from Synapse.

Users reported the audio player not working from our domain. It was a CORS issue; the necessary headers are now being sent from the media server, so playing media files locally will function correctly.

Your files would have played correctly via other instances.

@LovelessLiddell Thanks for letting me know. I patched the notification system to be compatible with the new schema until we get the 'follow user' feature which is what broke API access to your notifications. This is in line with issues I'd expect to see given the situation.

Notification functionality should be restored and working properly now. Have fun.

The resolution was bumping the code to v3.2.1 release, which appears to not have caused any ill effects. We may have to rollback if there are any significant issues, but the release was only 18 days ago.

Sorry, the server was mistakenly upgraded to bleeding edge which has a bug relevant to broken notifications. We may be riding this until there is another release.

I pulled a fix for the notification issue. Contact me know if it is not resolved.

The server ran out of space due to backups. Backup frequency has been temporarily reduced. The admin team will remediate this properly soon. Sorry for the inconvenience.

Apologies for the outage earlier today. The ruby process was getting killed by oomkiller.

We've since upgraded the server to double the available RAM.

Server is not equipped for global relaying, apologies for the latency you were experiencing.

Sorry for the hiccup, the server ran low on RAM. We may need to upgrade to accommodate the upload size that Mastodon demands.


A server, for mew.