>
Jump AcademyThe uptime issue for the EU database node is not caused by server performance but by desynchronization from the rest of the cluster. I have yet to automate the detection and automatic restart of disconnected nodes that failed to resynchronize with the cluster.
The join lag is caused by the clientprefs extension. Not only does it appear to not properly thread its queries, but it also
absolutely shits itself when it loses connection to the remote database. I'll have to migrate our clientprefs setup back to using local sqlite and set up a way to synchronize the settings across servers after I finish work for the Tip of the Hats event this weekend.
In the meantime, you will just have to notify me on Discord when it happens and I'll restart the corresponding database node manually.