FrostyX
Tracer doesn't understand that services are restarted
Originally reported here: https://bugzilla.redhat.com/show_bug.cgi?id=2275908
I know that there are some plans to replace the plugin by new functionality, so I'm not specifically asking for this to be fixed. I'm putting up this bug so that users who are confused can find it. The cloud image includes the plugin by default, so it's annoying that it returns bogus results.
$ sudo dnf reinstall systemd ... You should restart:
- Some applications using: sudo systemctl restart systemd-homed sudo systemctl restart systemd-journald sudo systemctl restart systemd-logind sudo systemctl restart systemd-oomd sudo systemctl restart systemd-resolved sudo systemctl restart systemd-udevd sudo systemctl restart systemd-userdbd
- These applications manually: (sd-pam) Additionally, there are:
- 1 processes requiring reboot For more information run: sudo tracer -iat 1713437258.5820656 Reinstalled: systemd-255.4-1.fc41.x86_64
But if you check with 'systemctl status', systemd-journald and systemd-userdbd were just restarted. (systemd-oomd, systemd-udevd, systemd-resolved are from different subpackages, and systemd-logind has a bug which does not allow it to be restarted).
Maybe this is caused by https://fedoraproject.org/wiki/Changes/Restart_services_at_end_of_rpm_transaction? But note that the implementation in the end is different than what we describe in the proposal. We mark units for restart or reload with '/usr/lib/systemd/systemd-update-helper mark-restart-system-units' and systemd has a %transfiletriggerpostun to execute the restarts at the end.
This is in Fedora-Cloud-Base-Generic.x86_64-Rawhide-20240415.n.0.qcow2: python3-dnf-plugin-tracer-4.1.2-1.fc40.noarch
Reproducible: Always
To clarify: systemd-logind is currently not restarted because of … reasons. But all the other systemd-* services are restarted upon package upgrades, so if they show up in the output, then most likely something's wrong with the test.
