When the provisioner calls rbd, if the configuration is incorrect, the command hangs and no logs are generated. Without logging into the container, it is difficult to determine what is going on.
It may be helpful if the exec could timeout and report the failure rather than simply hang.
Similarly, at a higher verbosity level, it would be helpful to print the full rbd command so that it could easily be tested at the rbd container command line when there are problems.
I can take this one.
Most helpful comment
I can take this one.