Unicorn takes 100% cpu and can't access forum

(zh99998) #1

I just migrate a forum to discourse today. It works well hours before, but it went read-only mode, then can’t visit minutes later. Seems auto-backuping.then about 10 mins later, seems backup is finished, the ruby, postgre, gzip process is not take much resources. But unicorn process began to take 100% cpu, and still can’t visit forum.

there is lots of

Kill self supervisor is gone

in unicorn.stdout.log
and lots of

E, [2016-01-15T13:09:21.547869 #22994] ERROR – : worker=1 PID:23084 timeout (5168s > 30s), killing

in unicorn.stderr.log

(zh99998) #2

I tried

RAILS_ENV=production rails s

it’s not responding.


docker kill app

not works.

I restarted machine and it works, but I still don’t know what happened.

(Kane York) #3

This is probably the aufs kernel bug where a SIGKILL during certain parts of a filesystem call (read/write) will hang the system call.