IBM ServeRAID, User Reference

Page: 178 / 226

v Restore or rebuild the data.
7. Restore or rebuild the missing data on each critical logical drive if any of the
following problems persists:
v The critical logical drive remains inaccessible.
v
Data errors are found on the critical logical drives.
v The operating system continuously fails to start properly.
v
Partition information on the critical logical drives is unknown.
Finishing the procedure
Complete the following steps:
1. Initiate a backup of the data.
2. After the backup is completed, replace the physical drive that was removed in
step 7 on page 161. The new physical drive must be equal to or greater in size
than the original drive. An auto-rebuild operation will start.
3. If necessary, redefine the hot-spare drives.
4. Using DUMPLOG, capture another set of ServeRAID logs.
5. Clear the ServeRAID logs using the following CLEARLOG.BAT command
available on the DUMPLOG Web site:
CLEARLOG controller
where controller is the ServeRAID controller number (1-12).
6. If a case was opened with your IBM service representative for this problem,
complete steps 7 through 9; otherwise, you have completed the recovery
process.
7. Using DUMPLOG, capture the ServeRAID logs after two to three days of normal
activity.
8. E-mail the ServeRAID logs to your IBM service representative.
9. If IBM determines that the ServeRAID controller and SCSI bus activities are not
operating properly, your IBM service representative will provide an action plan.
Follow the action plan, and repeat steps 7 and 8 until the system is running
normally.
Troubleshooting
If you continue to experience problems, review the following information. It might
help you identify the configuration or hardware problem.
Poor signal quality across the SCSI bus
The most common cause of multiple physical drive failures is poor signal quality
across the SCSI bus. As the SCSI protocol tries to recover, the overhead increases.
As the system becomes busier and demand for data increases, the corrective
actions of the SCSI protocol increase, and the SCSI bus moves toward saturation.
This overhead will eventually limit the normal device communication bandwidth. If
left unchecked, one or more SCSI physical drives might not respond to the
ServeRAID controller in a timely way, which will result in the ServeRAID controller
marking the physical drives defunct.
Poor signal quality problems can be caused by any of the following:
v Improper installation of the ServeRAID controller in a PCI slot
v Poor cable connections
v Poor seating of hot-swap drives against the SCSI backplane
v
Improper installation or seating of backplane repeaters
162
IBM ServeRAID: User’s Reference, Version 5.10