Placement endpoints not being updated, or perhaps regressing to n-c-c endpoints, even after deploying placement service for train
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Charms Deployment Guide |
Fix Released
|
High
|
Peter Matulis | ||
OpenStack Nova Cloud Controller Charm |
Fix Released
|
High
|
Billy Olsen | ||
OpenStack Placement Charm |
Invalid
|
High
|
Unassigned |
Bug Description
I recently upgraded a customer cloud from OpenStack Stein to Train, following the charm deployment guide for OpenStack: https:/
At this point, all O7k apps have been upgraded to Train, and we are planning to upgrade to Ussuri. However, we've encountered an issue. We've noted that the placement endpoitns listed via "openstack endpoint list" are using the nova-cloud-
When I first encountered this, I noted that I had not yet set up certificates for the placement service, so I added a vault to placement relation, after which the endpoints got updated. That was one or two days ago. However, when I checked again today, the endpoints again are using the nova-cc FQDN.
We're also having errors creating and deleting instances, which I'm suspecting may also be related to this.
Curl tests against the nova-cc FQDN on port 8778 fail with "connection refused" errors, as I would expect, while curl tests against the nova-placement FQDN on port 8778 succeeds. Yet, the endpoint list shows the nova-cc FQDN.
tags: | added: openstack-upgrade |
Changed in charm-placement: | |
status: | New → Confirmed |
status: | Confirmed → Triaged |
importance: | Undecided → High |
Changed in charm-deployment-guide: | |
status: | New → Confirmed |
importance: | Undecided → High |
Changed in charm-deployment-guide: | |
assignee: | nobody → Peter Matulis (petermatulis) |
status: | Confirmed → In Progress |
Changed in charm-deployment-guide: | |
status: | In Progress → Fix Released |
Changed in charm-nova-cloud-controller: | |
milestone: | none → 21.10 |
Changed in charm-nova-cloud-controller: | |
status: | Fix Committed → Fix Released |
I've manually adjusted the placement endpoints via "openstack endpoint set <id> --url <url>".
And while the instance creation issues may or may not be caused by this bug, instance creation is still not resolved, even with the above workaround.