Restart on suspend/resume/network change
On these events we should consider restarting to re-profile.
I've noticed several times now that systemctl status gives output "degraded" on boot because of an issue with the mirrorsorcerer service.
systemctl list-units --state=failed
UNIT LOAD ACTIVE SUB DESCRIPTION
mirrorsorcerer.service loaded failed failed Mirror Sorcerer
After restarting the service, there are no issues. Whatever the issue is doesn't seem to affect the functionality of pointing to faster repos.
Hmmm interesting. Do you have more logs from when it failed?
It may not be as important soon - zypper has finally gained geoip support so we will soon begin to phase out mirrorsorcerer singe zypper now does it internally.
Oh, very cool. And yes actually, I found some from logs I'd saved for something else -
28.10.2022 06:12:23:158 systemd Started Mirror Sorcerer ✨ 🪄 🪞 ✨.
28.10.2022 06:12:23:160 systemd Started NoMachine Server daemon.
28.10.2022 06:12:23:161 systemd Starting System Logging Service...
28.10.2022 06:12:23:229 mirrorsorcerer Mirror Sorcerer 🪄 🪞 ✨
28.10.2022 06:12:23:229 mirrorsorcerer irrorcache.opensuse.org to an ip address.
28.10.2022 06:12:23:229 mirrorsorcerer irrorcache-au.opensuse.org to an ip address.
28.10.2022 06:12:23:230 mirrorsorcerer irrorcache-us.opensuse.org to an ip address.
28.10.2022 06:12:23:230 mirrorsorcerer irrorcache-jp.opensuse.org to an ip address.
28.10.2022 06:12:23:230 mirrorsorcerer irrorcache-us-east.opensuse.org to an ip address.
28.10.2022 06:12:23:230 mirrorsorcerer irrorcache-us-west.opensuse.org to an ip address.
28.10.2022 06:12:23:230 mirrorsorcerer irrorcache-br.opensuse.org to an ip address.
28.10.2022 06:12:23:230 mirrorsorcerer Mirror profiling failed!
28.10.2022 06:12:23:231 systemd mirrorsorcerer.service: Main process exited, code=exited, status=1/FAILURE
28.10.2022 06:12:23:231 systemd mirrorsorcerer.service: Failed with result 'exit-code'.
I have a couple other logs from other days which show a result identical to the above, so it happens consistently.
Yeah, looks like networking isn't available, so it can't profile, and thus won't change anything, but that explains the failure. We could be more graceful in that scenario I guess.
@jmcelroy01 Okay, I'm going to test an update which improves the .service file to handle this better :)