20.08 ussuri ovn-chassis using the wrong hostname

[Márton Kiss]

A critical alert had been raised in nagios, because ovn-controller service had a wrong keepalive status in neutron agent list. This alert started to show up after a server restart required by sysconfig charm.

$ openstack network agent list | grep juju
| juju-0833f7-8-lxd-8.<tld-domain> | OVN Controller agent | juju-0833f7-8-lxd-8.<tld-domain> | | :-) | UP | ovn-controller |
| juju-0833f7-4-lxd-9.<tld-domain> | OVN Controller agent | juju-0833f7-4-lxd-9 | | XXX | UP | ovn-controller |
| juju-0833f7-6-lxd-7.<tld-domain> | OVN Controller agent | juju-0833f7-6-lxd-7.<tld-domain> | | :-) | UP | ovn-controller

The octavia/0 is using a short hostname instead of fqdn, and the ovn certificate had been issued using the shortname as well:

$ juju run --application octavia 'sudo cat /etc/ovn/cert_host | openssl x509 -text | grep Subject:'
- Stdout: |2
            Subject: CN = juju-0833f7-4-lxd-9
  UnitId: octavia/0
- Stdout: |2
            Subject: CN = juju-0833f7-6-lxd-7.<tld-domain>
  UnitId: octavia/1
- Stdout: |2
            Subject: CN = juju-0833f7-8-lxd-8.<tld-domain>
  UnitId: octavia/2

Removing the ovn-central:ovsdb octavia-ovn-chassis:ovsdb relation did not solved the issue, because the charm was running into error status