There was a test failure due to a pam_nologin failure when issuing a command over ssh. Curiously, there seemed to have been multiple successful ssh commands since the last reboot prior to the failure. A good start might be to enable the existing debug flag. Some details that would be useful understanding the failure:

• What login name does the command get sent as?
• Emitting a message on each ssh attempt and retry
• Assert that a pam_nologin error is something the ssh code retries on.

Additionally, if there's a way to track the changes to the pam_nologin file, that might uncover some false assumptions.