Rebuilding A Defunct Drive - IBM ServeRAID User Reference

Rebuilding a defunct drive

A physical drive is marked defunct when there is a loss of communication between
the controller and the physical drive. This can be caused by any of the following:
v An improperly connected cable, physical drive, or controller
v Loss of power to a drive
v An improperly assembled SCSI channel in an unsupported configuration
v A defective cable, backplane, physical drive, or controller
v Connecting unsupported SCSI devices (such as tape drives or CD-ROM drives)
In each case, after the communication problem is resolved, a rebuild operation is
required to reconstruct the data for the device in its disk array. The ServeRAID
controllers can reconstruct RAID level-1, level-1E, level-5, level-5E, level-10,
level-1E0, and level-50 logical drives. They cannot, however, reconstruct data
stored in RAID level-0 logical drives because RAID level-0 is not redundant. If an
array contains only RAID level-0 logical drives, the logical drives in the array are
marked offline and the logical drives contain damaged data. You cannot rebuild the
logical drives. You must correct the cause of the failure or replace the physical
drives; then, you must restore your data.
To prevent data-integrity problems, the ServeRAID controllers set the RAID level-0
logical drives in the affected array to blocked during the rebuild operation for RAID
level-1 or RAID level-5. After the rebuild operation is completed, you can unblock
the RAID level-0 logical drives and access them once again. Remember, however,
that the RAID level-0 logical drives contain damaged data.
Note: For logical drives in an IBM ServeRAID Cluster Solution:
Recovering from defunct drives
If the defunct drives are not part of an array, contact your IBM service
representative.
If a physical drive fails in an array or multiple physical drives fail in separate arrays
(one physical drive per array), complete the following steps:
1. Replace each defunct physical drive. The ServeRAID controller starts the
After the format process is complete, the ServeRAID controller will be able to
recognize the drive again.
to the same SCSI channel used for an array
v Because shared logical drives can have only one logical drive for each
array, blocking a RAID level-0 logical drive during a rebuild operation does
not apply to shared logical drives.
v Because non-shared logical drives can have more than one logical drive
for each array, blocking a RAID level-0 logical drive during a rebuild
operation does apply to non-shared logical drives.
rebuild operation when it detects the removal and reinsertion of a drive that is
part of an array.
Chapter 18. Solving ServeRAID problems
157