Compaq 3200 - Manual (Page 182)

H-2 Recovering from Drive Failure
Writer: Rickard Project: Compaq Smart Array 3200 Controller Reference Guide Comments: 340862-002
File Name: O-APPH.DOC Last Saved On: 12/7/98 1:47 PM
COMPAQ CONFIDENTIAL - NEED TO KNOW REQUIRED
Fault Tolerance and Drive Failure
In the event of a drive failure, the condition of the logical drive varies,
depending upon the fault-tolerance method used. Since a single array of physical
drives can contain multiple logical drives with different fault-tolerance methods,
conditions of each logical drive on the same array are not necessarily the same.
At any one time, if more drives are failed than the fault-tolerance method
allows, fault tolerance is “compromised” and the condition of the logical drive
can be referred to as “failed.” If a logical volume is “failed,” all requests from
the operating system will be rejected with “unrecoverable” errors.
A Non-Fault-Tolerant (RAID 0) Logical Drive
A Non-Fault-Tolerant (RAID 0) logical drive can sustain no drive failures. If
any physical drive in the array is failed, the condition of all non-fault-tolerant
logical drives in the same array will also be “failed” because data is striped
across all drives in the array.
A RAID 1 (Mirroring) Logical Drive
■ Can sustain multiple drive failures as long as failed drives are not
mirrored to one another
■ Will be in a “failed” condition if any two failed drives are mirrored to
one another
■ Will be in a “regenerating” condition if at least one drive is failed and no
failed drives are mirrored to one another
■ Will be in a “rebuilding” condition if a previously failed drive has been
replaced and the replacement drive is rebuilding. The volume may also
be in a “rebuilding” condition following a drive failure if a spare drive
was previously assigned and is being rebuilt.
In any RAID 1 logical drive on the Smart Array 3200 Controller tht has N
physical drives in the array (not including spare drives), the first N/2 physical
drives are normally mirrored in consecutive order to the second N/2 physical
drives. When determining the order, number each drive in the array by
increasing IDs on the first SCSI bus, followed by drives on the second SCSI
bus.