Do not restart condor on factory upgrade
GlideinWSM includes some configuration files for HTCondor (installed in /etc/condor/config.d). These include setting the daemons to be started and include shared port daemon configurations. These changes require a restart of HTCondor.
Currently we do automatically a condrestart in the post-install part of the RPM (specifically for usercollector, userschedd and factory).
This causes a restart of condor after the initial install and all the upgrades (even if not all upgrades change the daemons or the shared port setup).
This was enforced after a request a couple of years ago. [#13068]
Factory operations requested to change this in the factory. Also because the restart of condor could fail and it is not triggered manually, the failure may remain unnoticed.
Possible actions are:
1. Automatically trigger a condrestart (current one)
2. Automatically trigger a reload (will take care of most configuration changes but not when daemon list and shared port are involved)
3. Do nothing automatically. Warn the sysadmin that a restart of condor is needed
It seems that the preference (by Factory and Frontend operators queried via email) is for a reload (condor_reconfig) and the addition in the RPM installation document and in the release notes (of versions that need it) of requests to do a restart.