I have an installation of discourse that started responding with 504’s for every request this afternoon. After digging into it a bit, I discovered that unicorn was shutting down child workers after timeouts.
I’ve tried a number of things: rebuilding the container, restarting the host machine and the database server, and checking every log that I can find, but I’ve been unable to locate anything that seems like the source of the issue. I turned the
log_level for the Ruby app all the way up to
:debug, but all I can see are the timings for SQL statements (which seem fine).
What else can I do to diagnose this issue?