Reg same lun numbers on 3 nodes..

[chindi]

hi guys,

I have a query, is it necessary that the lun numbers of a vdisk which are presented to 3 nodes of a serviceguard cluster should be same...

what if i have a case like..primary node havng lun number 24 wheres secndry and tertiary node as lun number 26...

Will it create any problem when my package shifts from primary node due to node failure...

thanks in advance...

[Matti_Kurkela · Accepted solution]

Serviceguard does not require making the LUN numbers the same in all the nodes. Of course, if the numbers are the same, it's very slightly easier to verify that all the nodes are configured correctly.

If you use LVM volume groups in your Serviceguard package(s), when you create the map file for your package VG using "vgexport -s -p", the map file will include the VGID of the volume group. When importing that volume group to another node using "vgimport -s", the vgimport command will scan all the LUNs visible to the node and will automatically pick the correct LUNs by checking the VGID. The result is that on each node, the LVM subsystem knows the correct LUNs for that node; Serviceguard does not need to know or care about LUN numbers at all.

If you use VxVM instead, it has a different procedure but the basic idea is similar.

Of course, if you have a strange application that wants to use raw disks (not filesystems) *and* insists to use LUN numbers to identify the disks, then you would either make the LUN numbers the same in all the nodes, or write a special start-up script that changes the application configuration based on which node it's currently running. But applications that use raw disks are somewhat rare (mainly highly performance-optimized databases), and applications that use LUN numbers instead of device names are even rarer.

MK

[chindi]

Thanks Matti you were spot on....

[Basheer_2]

Yes, in my case 11.16 when we set up the cluster we always have the same LUN #s in all the nodes.