Oracle Solaris Cluster 3.3, Системное руководство

Страница: 38 / 56

(StorEdge A3500 Only) On both nodes, to prevent LUNs from automatic assignment to the
controller that is being brought online, set the
System_LunReDistribution
parameter in the
/etc/raid/rmparams
file to
false.
Caution –
You must set the
System_LunReDistribution
parameter in the
/etc/raid/rmparams
file to
false
so that no LUNs are assigned to the controller being brought online. After you
verify in Step 8 that the controller has the correct SCSI reservation state, you can balance LUNs
between both controllers.
For the procedure about how to modify the
rmparams
file, see the Sun StorEdge RAID Manager
Installation and Support Guide .
Restart the RAID Manager daemon.
# /etc/init.d/amdemon stop
# /etc/init.d/amdemon start
If your controller module is offline, but does not have a failed controller, proceed to Step 7 .
If you have a failed controller, replace the failed controller with a new controller. Do not bring the
controller online .
For the procedure about how to replace controllers, see the Sun StorEdge A3500/A3500FC
Controller Module Guide and the Sun StorEdge RAID Manager Installation and Support Guide
for additional considerations.
On one node, use the RAID Manager GUI's Recovery application to restore the controller online.
Note –
You must use the RAID Manager GUI's Recovery application to bring the controller
online. Do not use the Redundant Disk Array Controller Utility (
rdacutil
) because this utility
ignores the value of the
System_LunReDistribution
parameter in the
/etc/raid/rmparams
file.
For information about the Recovery application, see the Sun StorEdge RAID Manager User’s
Guide . If you have problems with bringing the controller online, see the Sun StorEdge RAID
Manager Installation and Support Guide .
On one node that is connected to the storage array or storage system, verify that the controller
has the correct SCSI reservation state.
Use the following command on LUN 0 of the controller you want to bring online.
In the following command, devicename is the full UNIX path name of the device, for example
/dev/dsk/c
1
t
X
d
Y
# cldevice repair
devicename
3
4
5
6
7
8
Maintaining Storage Arrays
Oracle Solaris Cluster 3.3 With StorEdge A1000 Array, Netra st A1000 Array, or StorEdge A3500 System Manual
• May 2011, Revision A
38