DirectAdmin stopped responding and will no longer come up with systemd

[adam12]

DirectAdmin wasn't responding to port :2222. I went into the system and see directadmin running and listening. Checked task queue and noticed it had a lot of commands that hadn't been run. I relocated the queue to see if it was the culprit and it didn't improve things.

Attempting to restart directadmin via systemctl failed or would timeout. I manually killed all processes relating to directadmin, and reset systemd, but starting directadmin.service still fails. I do not see any output in either journald or /var/log.

Here's the kicker: running directadmin by hand works fine. I crank up debug to 2000 and nothing looks out of place. But once I attempt to start it with systemd again, it goes unresponsive.

Running strace on the primary pid shows that it's waiting on a FUTEX, which is similar to how it normally runs, but when booting with systemd, there are no child pids to strace (which it explains why it's not listening on :2222).

I'm going to try to reboot tonight and see if there's some sort of weird state persisting.

DirectAdmin v.1.658 55d62056090f3abd3b2785330d708a0916ff8ab2

Has anyone else seen this?

 

[adam12]

Reboot seemed to help.

In case anyone else stumbles upon this, I found the service timing out when systemd was attempting to start it. Increasing StartTimeout to some absurd value didn't seem to help. The process appeared running but it wasn't accepting connections, so maybe there was some sort of deadlock somewhere.

 

[ccto]

In the past, in a few circumstance, we encountered similar case for Apache - "Couldn't create the rewrite-map mutex"
not sure it relates to your case or not.

We run the following to clear semaphore

service httpd stop
ipcs -s | grep apache | awk ' { print $2 } ' | xargs ipcrm sem
service httpd start

 

[adam12]

Thanks for the reply.

I've run into that issue before, and it's different in this case.