IBM ServeRAID-8 Series, Best Practices And Maintenance Information

Page: 8 / 58

Initializing previously used good hard disk drives
A previously used good hard disk drive is defined in this document as a drive that
was previously a member of an array that was attached to a ServeRAID-8 series
controller. The drive is being reused within a new RAID configuration. Before you
can add a previously used good hard disk drive to a new array, either as a
replacement drive or to expand an array, you must first initialize the drive to remove
existing configuration information. Existing configuration information can cause the
ServeRAID controller to behave differently than expected and, in some rare cases,
can result in data loss.
You can initialize a hard disk drive by using the Array Configuration Utility (ACU)
(accessible by pressing Ctrl+A when you are prompted at system startup), the IBM
ServeRAID Support CD version 8 or 9, or the ServeRAID Manager program (from
within the operating system), or by using the
arcconf task command.
For more information, see the following RETAIN tips.
A rebuild does not start after replacing a defunct drive - IBM ServeRAID
http://www-304.ibm.com/systems/support/supportsite.wss/
docdisplay?brandind=5000008&lndocid=MIGR-5074510
(Type
5074510 in the Search field at http://www.ibm.com.)
Lost configuration when drive added - ServeRAID 8k, 8k-l
http://www-304.ibm.com/systems/support/supportsite.wss/
docdisplay?brandind=5000008&lndocid=MIGR-5073723
(Type
5073723 in the Search field at http://www.ibm.com.)
Synchronizing logical drives after upgrading firmware from build 8263
or earlier
ServeRAID firmware builds 7777 through 8263 use a legacy method to track
bad-stripe unit errors. Starting with firmware build 8264 and later, the process to
manage bad-stripe units is improved with a bad-stripe table. The firmware upgrade
process does not import existing bad-stripe information from the old method to the
new, and the legacy method is perceived by the new code as disk errors when the
affected stripe units are read. In most cases, these types of errors are corrected
automatically without any user intervention. However, in rare cases, if a large
number of these errors are reported by a single drive, the drive might be marked
Defunct prematurely.
To avoid this situation, start a synchronization on each logical drive. The
synchronization process scrubs the physical drives, corrects the errors as they are
detected, and when applicable, creates an equally equivalent bad-stripe table entry.
Note that the ServeRAID-8 series controllers are designed to correct or self-heal
bad-stripe table entries when the next successful write occurs to that stripe unit.
You can start synchronizations by using the IBM ServeRAID Support CD, the
ServeRAID Manager installable application (from within the operating system), or
the
arcconf task command.
2
ServeRAID-8 Series: Best Practices and Maintenance Information