Comment 4 for bug 1407616

It was not a dependency issue.
After re-trying the provisioning from the VNC console, all slaves were provisioned smoothly on the affected env.

Looks like some connection issue to the master node happened during provisioning.