There are two kinds of RAID malfunctions users can experience. One is when less disks fail that are required to rebuild an array, second is when more disks are damaged (also giving I/O errors). For example 2 or more disks in RAID 5, 3 or more disks in RAID 6 etc.
RAID degraded, but rebuildable:
All data on RAID is still accessible, although due to the rebuild process system performance can be compromised. In older versions of DSS V6 (5377 and lower) it is advised to switch active node (if failure appeared on primary) to secondary for time of rebuild (also for data safety in case another disk in rebuilding array goes out). After rebuild is complete synchronize volumes from secondary node and failback.
RAID damaged, unrebuildable:
New RAID must be created. In this case it is not possible to synchronize volumes because replication tasks will not recognize disks UIDs. Therefore data must be replicated manually and new failover tasks configured:
- Deactivate failover functionality
- Remove old replication tasks
- Replicate data to newly created RAID
- Configure and start new failover
On more informations on how to add clean node to failover please refer to this KB article:
kb.open-e.com/How-can-we-setup-the-primary-node-restart-the-iSCSI-Failover-and-make-this-node-active-again_1053.html
