Comment 7 for bug 1878248

For stable/train random failures might be related to the bug:
https://github.com/cirros-dev/cirros/issues/8

While we spawn more VMs during the test we found that sometimes cirros can't get public ssh keys and fails without retrying.
It ends with test failure that the authentication failed.

I added the retry to Cirros ec2metadata script and it has been recently released. We should try to change Cirros to 0.5.1 for devstack and in tripleo tests in order to prevent this situation to happen again.

https://github.com/cirros-dev/cirros/tree/0.5.1

Taken from stable/train failure:
2020-05-20 09:25:23,053 346350 WARNING [tempest.lib.common.ssh] Failed to establish authenticated ssh connection to cirros@10.0.0.115 (Authentication failed.). Number attempts: 21. Retry after 22 seconds.
2020-05-20 09:25:45,598 346350 INFO [paramiko.transport] Connected (version 2.0, client dropbear_2015.67)
2020-05-20 09:25:45,841 346350 INFO [paramiko.transport] Authentication (publickey) failed.
2020-05-20 09:25:45,973 346350 WARNING [tempest.lib.common.ssh] Failed to establish authenticated ssh connection to cirros@10.0.0.115 (Authentication failed.). Number attempts: 22. Retry after 23 seconds.