Failed service not triggering rollback

[slaecker]

On MicroOS I recently observed that k3s fails on kernel 6.6.58-1-longterm but starts up successfully on kernel-default. So I wanted to add it to the health-checker to force a rollback if a kernel update breaks k3s.

I added k3s.service to the After section of health-checker.service but the health-check always succeeded.

cat /etc/systemd/system/health-checker.service.d/override.conf
[Unit]
After=k3s.service

Because k3s.service stayed in status Activating due to Restart=always I added an override:

cat /etc/systemd/system/k3s.service.d/override.conf
[Service]
Restart=on-success

Now when rebooting the service fails but health-checker won't perform a rollback. The check still succeeds.

Am I missing something?

systemctl status k3s.service 
× k3s.service - Lightweight Kubernetes
     Loaded: loaded (/etc/systemd/system/k3s.service; enabled; preset: disabled)
    Drop-In: /etc/systemd/system/k3s.service.d
             └─override.conf
     Active: failed (Result: exit-code) since Fri 2024-11-01 12:14:25 CET; 12min ago
   Duration: 1.254s
 Invocation: 3435ccb41df54841b4e1482cc86e52d4
       Docs: https://k3s.io
   Main PID: 1163 (code=exited, status=2)
        CPU: 6.846s

systemctl status health-checker.service 
○ health-checker.service - MicroOS Health Checker
     Loaded: loaded (/usr/lib/systemd/system/health-checker.service; enabled; preset: enabled)
    Drop-In: /etc/systemd/system/health-checker.service.d
             └─override.conf
     Active: inactive (dead) since Fri 2024-11-01 12:14:25 CET; 13min ago
 Invocation: 30a45f6a1e1645f584fd606313f48934
   Main PID: 1326 (code=exited, status=0/SUCCESS)
        CPU: 620ms

Nov 01 12:14:24 vm-opensuse-microos-test systemd[1]: Starting MicroOS Health Checker...
Nov 01 12:14:24 vm-opensuse-microos-test health-checker[1326]: Clearing GRUB flag
Nov 01 12:14:24 vm-opensuse-microos-test health-checker[1327]: grub2-editenv: error: cannot open `/boot/grub2/grubenv': Read-only file system.
Nov 01 12:14:24 vm-opensuse-microos-test health-checker[1326]: Starting health check
Nov 01 12:14:24 vm-opensuse-microos-test health-checker[1373]: active
Nov 01 12:14:25 vm-opensuse-microos-test health-checker[1326]: Health check passed

[thkukuk]

There is no plugin for k3s, so health-checker cannot verify that and ignores the service.
Somebody needs to write and provide a plugin first.

[slaecker]

Thanks, I worked on this today and created pull request #27