Determining A Controller Failure - Sun Microsystems Sun StorEdge A3500 Manual

6.3

Determining a Controller Failure

If there is a problem with the controller, you will see two amber LEDs glowing on
the controller module: the controller fault LED on the front panel and a fault LED on
the failed controller canister. There are four circumstances that can cause a controller
fault:
Controller failure
SIMMs module failure (processor or cache memory)
Controller back panel failure
Controller overheating
Before you replace a failed controller, be aware of the following:
In the event of a controller failure, the other controller will take over and assume
task processing functions. However, you will probably lose any data stored in the
cache memory of the failed controller. You should stop all I/O activity to the
failed controller to prevent more data loss or other communication problems on
the bus.
You can hot-swap a failed controller (replace it while the controller module is
operating) as long as the failed controller:
To ensure the best results, make sure that the new controller is the same as the
one you are replacing (same size memory, firmware version, and so forth).
Caution – Only remove a controller that has failed and has an amber Fault LED lit
!
or that you have marked as Failed (offline) through the RAID Manager software.
Removing a controller that is operating normally (not failed) will result in data loss.
1. Read through this entire procedure before performing any tasks.
a. Compare this information with the controller replacement and recovery
instructions in the Sun StorEdge RAID Manager 6.22 User's Guide.
b. Run "Recovery Guru" in the Sun StorEdge RAID Manager software.
c. Determine if you need to modify the remaining steps to accommodate
requirements unique to your software.
2. Check the amber controller LED on the front panel to see if there is a fault.
See Section 4.1 "Checking the Controller Module LEDs" on page 4-2.
6-4 Sun StorEdge A3500/A3500FC Controller Module Guide • August 1999
Is one of a redundant pair, two controllers attached to the same host, and
Has a passive status in the Sun StorEdge RAID Manager software.