action #89497
closed
flaky Failed systemd services alert (except openqa.suse.de)
Added by livdywan about 4 years ago.
Updated about 4 years ago.
Description
Observation¶
[Alerting] Failed systemd services alert (except openqa.suse.de)
Check failed systemd services on hosts with `systemctl --failed`. Hint: Go to parent dashboard https://stats.openqa-monitor.qa.suse.de/d/KToPYLEWz/failed-systemd-services to see a list of affected hosts.
Metric name
Value
Sum of failed systemd services
1.000
Alerting at 12:48Z, OK at 12.49Z.
Once more alerting at 13.03Z, OK at 13.07Z
Once more alerting at 13.27Z, OK at 13.37Z
Once more alerting at 13.49Z, OK at 13.52Z
Once more alerting at 2.30Z, OK at 2.31Z
Once more alerting at 2.37Z, OK at 2.40Z on openqaworker13
Suggestion¶
- Investigate how to include failed services in the alert notification (to make it clear if new alerts are flaky or new issues)
- Observe failed systemd services
- Check what
sudo salt -l error --no-color -C 'G@roles:worker' cmd.run "/etc/telegraf/scripts/systemd_failed.sh"
says
- Description updated (diff)
- Description updated (diff)
- Description updated (diff)
- Description updated (diff)
- Status changed from New to Feedback
It looks like the "flaky" alerts have been caused by manual reboots i.e. #88900 and the offending unit was var-lib-openqa-share.mount
- Status changed from Feedback to Workable
Status changed from Feedback to Workable
Can you please state why you changed this back to Workable?
There was no assignee and I would not know how we would ever look into tickets that are in "Feedback" but without assignee. If you consider that "Workable" sounds too much "workable", well then "New" would be ok for me :)
I observe 9 alerts Failed systemd services alert (except openqa.suse.de)
between March 6, 2.14Z and March 7, 11.05Z. Suspected to be due to automatic reboots after updates, alerting 2-3 minutes each.
- Description updated (diff)
- Status changed from Workable to In Progress
- Assignee set to nicksinger
- Due date set to 2021-03-31
Setting due date based on mean cycle time of SUSE QE Tools
- Status changed from In Progress to Feedback
I will set this to feedback to give it some more days to check if the "flakiness" is gone by bumping the "pending timeout" to 15m
- Status changed from Feedback to Resolved
I've seen the alert on April 7, where it was caused by the os-autoinst-openvswitch service on openqaworker-arm-1 and on April 1 where it was "expected" due to @okurz working on #81058 so I'd say this is not flaky at this point.
Also available in: Atom
PDF