Cinder's DRBD is left in "Inconsistent" state when a disk
replacement is attempted on AIO SX.
The fix enables DRBD cinder volumes reconfiguration in two cases:
1) At initial configuration,
2) When disk is replaced.
A disk replacement is triggered from sysinv by removing
the checkpoint file behind is_node_cinder_lvm.
On SX we have to wait for disk wipe confirmation before
updating DB otherwise user may unlock host without wipe
and operation won't be retried leading to major, possibly
unrecoverable, problems with cinder-volumes disk.
If DB was not updated, operation will be retried on reboot
ensuring that the disk is indeed wiped.
Change-Id: Ia3391b2437d309c71bb525b8167ac55d727b3ab2
19a756882b