G-4
Recovering from Drive Failure
Writer: Pamela King Project: SMART-2DH Array Controller Reference Guide Comments: 295469-002
File Name: N-APPG.DOC Last Saved On: 2/27/98 12:09 PM
COMPAQ CONFIDENTIAL - NEED TO KNOW REQUIRED
Replacing a Failed Drive
Failed drives in hot-pluggable trays can be removed and replaced while host
system and storage system power are both ON. Hot-pluggable drives can also
be replaced when the power is OFF. Remember, however, to NEVER TURN
OFF an external storage system while the host system power is on. This
would result in the failure of all drives in the storage system, which would
likely compromise your fault tolerance. When a hot-pluggable drive is
inserted, all disk activity on the controller will be temporarily paused while
the drive is spinning up (usually 20 seconds or so). If the drive is inserted
while power is ON, in fault-tolerant configurations, recovery of data on the
replacement drive will automatically begin (indicated by blinking online
LED).
Non-hot-pluggable drives should only be replaced while the system power is
OFF. On all non-hot-pluggable drives, be sure to check the SCSI ID jumpers
to make sure that the correct drive is being replaced. To prevent compromising
the fault tolerance, it is imperative that the correct physical drive be replaced.
Also, be sure that the SCSI ID jumpers are set to the same SCSI ID on the
replacement drive. Note that the SCSI ID jumpers may be located at different
places on different drive models, but it is imperative that the SCSI ID on the
replacement drive always be set to the same value as the original failed drive
to prevent SCSI ID conflicts which could compromise the fault tolerance.
The capacity of replacement drives must be at least as large as the capacity of
the other drives in the array. Drives that are insufficient capacity will
immediately be failed by the controller without starting the Automatic Data
Recovery.
CAUTION:
If the SMART-2DH Array Controller has a failed
drive, replace this drive with a new or known-good replacement
drive. In some cases, a drive that has previously failed by the
controller may appear to be operational after the system is
power-cycled, or after removal and reinsertion of a hot-pluggable
drive. However, this practice is highly discouraged as the use of
such “marginal” drives may eventually result in data loss.