SAS RAID controller maintenance

Ensure optimal performance of your controller by using these maintenance procedures.

To help avoid controller and disk array problems, use the following tips:

• Always perform a normal system shutdown before physically replacing or moving the RAID controller or members of disk arrays. A normal shutdown of the system will flush the controller's write cache and remove dependencies between the controller and the pdisks. Unconfiguring the controller by using the rmdev command (for example, rmdev -Rl sissas3) has the same effect as it would on a single controller when the system shutdown command is used.
  Note: pdisks that are a Failed member of a Degraded disk array can be replaced and the disk array reconstructed while the system continues to run. No system shutdown is required.
• You can physically move pdisks from one controller to another. However, if the pdisks are members of a disk array, be sure to move all the disks in the array as a group. Prior to attempting a disk movement, ensure that the disk array is not in a Degraded state because of a disk failure, and the controllers are unconfigured.
• When physically removing pdisks that are members of a disk array and there is no need to preserve data and no intent to use the disk array again, delete the disk array before removing the disks. This action avoids disk-array-related problems the next time that these disks are used.
• Always use the SCSI and SCSI RAID Hot Plug Manager to remove and replace a pdisk if performing a concurrent disk replacement. For instructions on how to remove and replace a disk, see Replacing pdisks.
• If a disk array is being used as a boot device and the system fails to boot because of a suspected disk array problem, boot using the Standalone Diagnostic media. Error Log Analysis, AIX® error logs, the IBM SAS Disk Array Manager, and other tools are available on the Standalone Diagnostics media to help determine and resolve the problem with the disk array.
• Do not attempt to correct problems by swapping controllers and disks unless you are directed to do so by the service procedures. Use Error Log Analysis to determine what actions to perform, and when appropriate, follow the appropriate MAPs for problem determination. If multiple errors occur at approximately the same time, look at them as a whole to determine if there is a common cause. For additional information regarding problem determination, see Problem Determination and Recovery.
• Do not confuse the cache directory card, which is a small rectangular card with round, button-shaped batteries, for a removable cache card. The nonvolatile write cache memory is integrated into the controller. The write cache memory itself is battery-backed by the large, rechargeable cache battery pack. The cache directory card contains only a secondary copy of the write cache directory and no cache data. Do not remove this card except under very specific recovery cases as described in the Maintenance Analysis Procedures (MAPs).
• Do not unplug or exchange a cache battery pack without following the procedures as outlined in this section or in the MAPs. Failure to follow these procedures might result in data loss.
• When invoking diagnostic routines for a controller, use the Problem Determination (PD) mode instead of System Verification (SV) mode unless there is a specific reason to use SV mode (for example, you were directed to run SV mode by a MAP).
• After diagnostic routines for a controller are run in SV mode, run diagnostics in PD mode to ensure that new errors are analyzed. These actions should be performed especially when using Standalone Diagnostic media.