[BUG] Compute nodes failing to run jobs because users_groups.json is empty

[cartalla]

Describe the bug

The compute node instance is launching, but slurmd can't launch the job because the admin1 user doesn't exist because users_groups.json is empty.

This is supposed to be created on the RES cluster manager instance, but it is no longer joined to the domain so the RES users and groups aren't available.