[BlueOnyx:21357] Re: Sendmail stopping on 5209R *** Update to CentOS 7.4 ***

[Michael Stauber]

Hi all,

> Not sure if anybody else is seeing this but all our 5209R VPS servers are having problems with sendmail stopping.
> 5207/5208R VPS are fine.
> Restart gets it going again but then it stops 30 secs later. Doesn't matter whether AVSpam is installed.
> 
> I have emailed Michael to see if he is able to help, but wondered if it was just me!.

I just checked it and can confirm the problem. It affected several (but
not all) of my 5209R as well after last nights YUM update to CentOS 7.4.

The problem seems to be that on some 5209R installs (but not all) the
symlink from /var/run to /run went missing and got replaced with a
regular directory. Sendmail seems to take offense on this.

If this happened to you, a "systemctl restart sendmail" will hang for a
long time and you eventually see this in /var/log/messages:


Sep 14 11:27:42 sol systemd: sendmail.service start operation timed out.
Terminating.
Sep 14 11:27:42 sol systemd: Failed to start Sendmail Mail Transport Agent.
Sep 14 11:27:42 sol systemd: Dependency failed for Sendmail Mail
Transport Client.
Sep 14 11:27:42 sol systemd: Job sm-client.service/start failed with
result 'dependency'.
Sep 14 11:27:42 sol systemd: Unit sendmail.service entered failed state.
Sep 14 11:27:42 sol systemd: sendmail.service failed.

One work around that works is this:

Edit /lib/systemd/system/sendmail.service and find this line:

PIDFile=/var/run/sendmail.pid

Change it to this:

PIDFile=/run/sendmail.pid

Then run this two commands:

systemctl daemon-reload
systemctl restart sendmail

The first notifies Systemd about the changed Sendmail unit-file and the
second one then successfully restart Sendmail.

I'm still looking at the broader issue of the /var/run symlink and why
that went away and what other issues (if any) it may have for us.

-- 
With best regards

Michael Stauber