RFE: Automated server upgrade

[JohnStrunk]

Given the number of servers we have, it takes a full day to upgrade everything manually.

We need a playbook that can perform the upgrade automatically, following the correct procedure.

There are a number of things that need to be done here:

• Role or playbook that can check and/or wait for:
  • All servers in a cluster to be up
  • All glusterd and bricks to be up
  • All volumes healed
  • No split-brain
• One host at a time, rolling upgrade
  • Choice between full or security-only updates
  • Automatically determine if node needs a reboot
  • gluster, kernel, or libc(?) update
• A streamlined version of the above needs to be available for the jump host, too