The mailing list for listmasters using Sympa

[David Carter <address@concealed>]

We have been asked to implement automated patching of all of our 
systems, including the system which runs Sympa on RHEL 9.  This would 
involve some patching and a possible reboot in the middle of the night 
when noone is around to fix problems, so the process has to be reliable.

When I patch this system by hand, I would normally shut down Sympa and 
the PostgreSQL database first. The full sequence looks something like:

systemctl stop httpd
systemctl stop exim
systemctl stop sympa
systemctl stop wwsympa (*)
systemctl stop sympasoap (*)
systemctl stop postgresql
dnf update

The complication is that the two "systemctl stop" command labelled (*) 
have a nasty habit of hanging: when I patch this system by hand I often 
have to ^C and rerun the "sytemctl stop" command or kill specific 
processes by hand. Obviously we can't automate that bit.

I have also tried:

systemctl stop wwsympa.socket
systemctl stop sympasoap.socket

That didn't seem to help.

Some experiments early this morning indicate that the problem is that 
systemd likes to restart wwsympa and sympasoap when it is supposed to be 
shutting them down. This might have something to do with:

/usr/lib/systemd/system/wwsympa.service
Restart=always
RestartSec=5

I guess that one approach would be to simply skip:

systemctl stop wwsympa
systemctl stop sympasoap

as those services should be isolated from the world so long as Apache 
httpd is shut down.

Can I ask what other people do to shut Sympa do?

Has anyone tweaked the

/usr/lib/systemd/system/wwsympa.service
/usr/lib/systemd/system/sympasoap.service

files which are provided by the Sympa rpm provided by EPEL so that:

systemctl stop wwsympa
systemctl stop sympasoap

work more reliably? Thanks in advance for any responses,

--
David Carter                             Email: address@concealed
University of Cambridge,
Information Services,
7 J J Thomson Avenue,
Cambridge UK.  CB3 0RB