Closed wallentx closed 2 months ago
Thanks, @wallentx ! I've always had a hard time understanding that, too. Also strange that a *Sec=
parameter would also accept minutes and other measurements of time.
Lol yeah, I guess it's probably safe to always have it there in case you forget to specify a unit in the value. Or when you aren't really sure what something like startLimitBurst
is.. and you just casually say "I'll take 3", and rest easy knowing that you'll get 3 somethings, rather than an error 😁
Good point! The name of the key at least gives users a hint of where to start by using seconds at least.
I just did some testing by spinning up a winesapOS 4.1.0-beta.0 VM, ripping out Mesa (sudo pacman -R -n -s -d -d mesa
), and then rebooting. It turns out that, despite sddm
throwing lots of errors, the service is still listed as "active (running)" so it never actually triggers the restart policy.
When I recently changed from LightDM to SDDM, I migrated all of the configurations over but forgot to test it. Perhaps I can make a small shell script to check for errors in the SDDM journal.
Here are some of the error messages:
Failed to read display number from pipe
Display server stopping...
Attempt 3 starting the Display server on vt 2 failed
Could not start Display server on vt2
I'll go ahead and merge this in since the SDDM failure recovery not actually working is a separate issue.
When I initially made these changes, I had some trouble finding good details on
StartLimitInterval
vsStartLimitIntervalSec
, and when/where to use it. I had a hunch that I had used the legacy syntax here, but I was able to actually confirm it just now from details in this PR https://github.com/systemd/systemd/issues/16644