We don't quite understand what is the connection between the Redis and not being able to launch VMs. It seems like ovn-dbs bundle is the reason for that.
If customer could reproduce the issue and restart only ovn-dbs bundle and see if that's solves the problem.
Also we would like to see redis logs before and after redis stopped running.

From the report it is not clear if redis and ovn-db were restarted in the same step? Redis is not used when instances are launched.
A network disruption could explain the issues described and seen here.

CloudOps does not maintain redis or pacemaker.