Do not skip these.
NEVER snapshot or quiesce the Grid Infrastructure clusterware disk group (OCR / voting files — GRID, CRS, or +OCR). It holds clusterware metadata, not database data, and is restored through normal Grid Infrastructure procedures. Only the database disk groups (DATA, FRA, REDO) belong in the Protection Group.
Always snapshot ALL database disk groups together, in one Protection Group, so the capture is write-order consistent.
Use an Everpure Protection Group or a snapshot listing all databases volumes; never snapshot individual volumes separately.
Verify Protection Group membership before the maintenance window starts.
For application-consistent snapshots, always release backup mode or resume the instance even if the snapshot fails. Use error traps so the database is never left in backup mode or suspended.