I suggest the following steps for Ceph system tests which should be done before each snapshot/revert:
Before to freeze VM sand make snapshots:
* stop all ceph services at all nodes (gracefully)
After the snapshots has been reverted and VMs unfreezed:
* Ensure time in sync
* Start all ceph services
* wait for ceph_health reports OK for all nodes.
I suggest the following steps for Ceph system tests which should be done before each snapshot/revert:
Before to freeze VM sand make snapshots:
* stop all ceph services at all nodes (gracefully)
After the snapshots has been reverted and VMs unfreezed:
* Ensure time in sync
* Start all ceph services
* wait for ceph_health reports OK for all nodes.