In check_nova_services there is some faulty logic that needs to be sorted.
Hosts being checked as "(not-part-of-any-agg)" (or binned there because they are in a skipped aggregate) are triggering aggregate-level-hosts-available alarms, but should only be checked for down/disabled.
If I either have a single host that is not in an aggregate, or I have an aggregate that is skipped, and there is only one or less active hosts in that skipped aggregate, I get a critical alert about not enough hosts in the aggregate are alive.
There should be some logic to skip the aggregate hosts count checks in check_hosts_up() if the aggregate name is "(not-part-of-any-agg)".
(by afreiberger)
This has been pushed to cs:~bootstack-charmers-next/openstack-service-checks and will be released in the next LMA release cycle expected in 21.02
In check_nova_services there is some faulty logic that needs to be sorted.
Hosts being checked as "(not-part-of-any-agg)" (or binned there because they are in a skipped aggregate) are triggering aggregate-level-hosts-available alarms, but should only be checked for down/disabled.
If I either have a single host that is not in an aggregate, or I have an aggregate that is skipped, and there is only one or less active hosts in that skipped aggregate, I get a critical alert about not enough hosts in the aggregate are alive.
There should be some logic to skip the aggregate hosts count checks in check_hosts_up() if the aggregate name is "(not-part-of-any-agg)".
Imported from Launchpad using lp2gh.
date created: 2020-12-08T15:50:01Z
owner: afreiberger
assignee: afreiberger
the launchpad url