Hey Folks,

It seems there are quite a few issues around Nova. I will investigate. To help me track, I am going to put everything here.

• Wrong hypervisor vCPUs calculation if node reports multiple NUMA cells #494
• openstack_nova_running_vms metrics erroring out #491
• openstack_nova_quota* metrics failed to collect due to caching issue #492
• OpenStack Exporter on RHEL 8.10 with OpenStack 17: "Resource not found" errors for security groups metrics #470
• Add nova/servergroups metrics #473
• Collecting hypervisor metrics from Nova fails when a new Ironic node is enrolled but not fully set up #428
• Lots of Nova metrics moved to Placement with Wallaby release (OS_COMPUTE_API >= 2.88) #249
• Failed to collect metric for exporter: nova #268
• openstack_nova_security_groups metric unavailable  #221
• openstack exporter nova_server_status fails to return flavor id #220
• Add project label for OpenStack metrics #526

My intuition says that a combination of something must have changed in Nova to cause all of these (#491 (comment)), along with changes we have made with 2nd Order Consequences we did not see (#466).

My hope is that the integration tests I am working on in #481 can be a solution to help avoid this. I also think it might be worth contacting https://github.com/canonical/charmed-openstack-exporter-snap to potentially de-sync master from snap if this is happening repeatedly.