5.5 HP StorageWorks X9720 Network Storage System Administrator Guide (AW549-96026, March 2011)

2. Immediately run the following command:
# exds_escalate
This gathers log information that is useful in diagnosing whether the data can be recovered.
Generally, if the failure is due to real disk failures, the data cannot be recovered. However, if
the failure is due to an inadvertent removal of a working disk drive, it may be possible to restore
the LUN to operation.
3. Contact HP Support as soon as possible.
Apparent failure of HP P700m
Sometimes when a server is booted, the HP P700m cannot access the SAS fabric. This is more common
when a new blade has just been inserted into the blade chassis, but can occur on other occasions.
Symptoms include:
• The HP P700m reports a POST error—this is visible using the TFT monitor/keyboard.
• The server crashes when the cciss driver loads— this is visible using the TFT monitor/keyboard.
Sometimes this happens to all servers in the system.
• No controllers are seen when you run the exds_stdiag command.
The underlying causes of these problems differ. However, the recovery process is similar in all cases.
Do not replace the HP P700m until you have worked through the process described here. In general
terms, the solution is to reset the SAS switches and if that fails, reboot each X9700c controller until
you locate a controller that is interfering with the SAS fabric.
If your system is in production, follow the steps below to minimize downtime on the system:
1. Log in to the Onboard Administrator and run the show bay info all command. Compare
entries for the affected blade and working blades.
If the entries look different, reboot each Onboard Administrator, one at a time.
Re-seat or replace the P700m in the affected server blade.
2. Run exds_stdiag. If exds_stdiag detects the same capacity blocks and X9720c controllers
as the other server blades, then the procedure is completed; otherwise, continue to the next step.
3. If all servers are affected, shut down all servers; if a subset of servers is affected, shut down the
subset.
4. Using OA, log into the SAS switch 1 and reset it.
5. Wait for it to reboot.
6. Reset SAS switch 2.
7. Wait for it to reboot.
8. Boot one affected server.
9. Run the following command:
# exds_stdiag
10. If X9700c controllers can be seen, boot other affected servers and run exds_stdiag on each.
If they also see the X9700c controllers, the procedure is completed; otherwise continue to the
next step.
Troubleshooting126