This chapter contains the following sections:
This chapter provides tips for troubleshooting a RAID system and describes possible solutions for common problems. If you are called to a site with a failed RAID subsystem or disk controller problem, you should check the following lists if the solution is not immediately obvious.
If a particular RAID array is off line but other disks on the same bus are functioning:
Two or more disks in the array may have failed.
The controller may be malfunctioning.
There could be a voltage problem on the RAID backplane.
There could be a voltage problem on the RAID backplane adapter.
There could be an unrecognized disk(s) in the RAID array.
The RAID box On/Off switch or power supply may have failed.
The physical cable connections and termination may be faulty.
If there are high rates of SCSI bus errors on the bus supporting the RAID(s):
Check the terminators if the bus is only supporting a single RAID box.
Make sure each SCSI disk in the RAID is a Silicon Graphics 2.0 GB IBM.
Check all cable connections and the backplane configuration in the box.
Verify that there are not duplicate SCSI IDs on the bus.
Verify that the 81-foot (25 m) differential SCSI bus length limitation has not been exceeded.
 | Note: Although all Vault-mounted RAID systems ship with a 25-foot (7.6 m) cable, non-RAID Vault expansion chassis may have a 50-foot (15.2 m) cable connecting them to the host system. |
If the RAID array has taken a disk off line because of a malfunction of some type, there are at least four ways to determine which disk it is. Most likely the disk has failed because of mechanical problems or an extremely high soft error rate. Chapter 1, "Introducing the RAID Hardware Options" lists the ways to positively identify a failed disk.
It is possible that the disk failed because the 5V or 12V supply feeding the disk assembly has failed. When you remove the failed disk during the hot-plugging procedure, check the two green voltage indication LEDs on the backplane where you will plug in the new disk. If one or both are dark, you probably have a voltage supply problem rather than a bad disk. The RAID box backplane should be replaced.
If you must service a Vault-mounted RAID box that will not power on, check the switch function before replacing the supply. A faulty switch can eliminate power even if the supply is in perfect working order.
Replacing the switch is a five-minute procedure that could save the bother and expense of replacing a power supply. You should carry a spare switch to every site that uses the Vault-mounted SCSI boxes.
You will need a small Phillips-head screwdriver and a pair of needle-nose pliers to quickly and safely remove the switch. Use the following information to replace a suspect switch;
Power off the Vault RAID SCSI box and unplug the power cord from the back.
Remove the two Phillips-head screws on the side of the switch housing.
Using the pliers, remove the two green and two brown connecting wires from the back of the switch. Be sure to note their positions for reference when the new switch is installed.
Press down on the bow retainers on the switch and pop it out the front of the housing.
Pop the new switch into the housing, making sure that the "o" shows on the top.
Reattach the four connecting wires to the back of the switch.
Resecure the switch housing to the side of the Vault box.
Reestablish power to the Vault box and try the new switch.