nova-cloud-controller charm doesn't seem to respect worker-multipler for nova-scheduler

[Brad Marshall]

I've deployed an Openstack cluster using Ubuntu 20.04 and Ussari with the 20.08 charms. This is deployed as a HA cluster, so 3 instances of each service across the servers.

Its been running for a number of months, and I've noticed it has a rather large number of nova-scheduler instances running for what I understand should be based on my config.

I've set worker-multipler to 0.0625 as the hardware presents as having 128 CPUs, and I have 128 nova-scheduler processes running, when I would have expected 8 based on the config.

The same value set for neutron-api gives me 8 running neutron-server process, as expected.

Please let me know if you need any further details.

Details:
OS: Ubuntu 20.04
Charm version: 20.08
nova-cloud-controller charm: jujucharms rev 346
neutron-api charm: jujucharms rev 287
juju controller version: 2.8.6

[Nobuto Murata (nobuto)]

Hi,

It sounds like a very similar issue we've fixed in 20.10 charm release:
https://bugs.launchpad.net/charm-nova-cloud-controller/+bug/1889756
Can you please take a look?

[Brad Marshall (brad-marshall)]

That does indeed look like the same thing, I've been looking at scheduling an upgrade of the charms on this system, I think this has just pushed the priority up a bit. Thanks for the quick response.