Reported at: https://bugzilla.redhat.com/show_bug.cgi?id=2008425
As part of the ovs2ovn migration process, some workload is created before the migration begins. The intention is to verify that those workloads are healthy after the migration.
The script that creates those workloads checks they are pingable and ssh'able. Sometimes the ssh fails because the created VMs did not get the public keys during the cloud-init phase:
+ ssh -i /home/stack/ovn_migration/pre_migration_resources/ovn_migration_ssh_key -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null cirros@10.0.0.243 date
Warning: Permanently added '10.0.0.243' (ECDSA) to the list of known hosts.
Permission denied, please try again.
Permission denied, please try again.
cirros@10.0.0.243: Permission denied (publickey,password).
+ rc=255
+ echo 'Done with the resource creation : exiting with 255'
Done with the resource creation : exiting with 255
+ exit 255
It seems the issue is related to a problem in cirros 0.4.0, which was resolved in a later release of cirros:
https://github.com/cirros-dev/cirros/pull/11/commits/e40bcd2964aa496a9d03e1aaf95cf7a86938f129
So, using cirros 0.5.2 could stabilize this script.
This script can be found in two different formats:
https://opendev.org/openstack/neutron/src/branch/master/tools/ovn_migration/tripleo_environment/playbooks/roles/resources/create/templates/create-resources.sh.j2
https://opendev.org/openstack/neutron/src/branch/master/tools/ovn_migration/infrared/tripleo-ovn-migration/roles/create-resources/templates/create-resources.sh.j2
Fix proposed to branch: master
Review: https:/