mcsg with rac need to migrate old to new array, steps ?

[Sammy_2]

HPUX 11.11
HP MCSG with RAC A11.16

I know, I need to modify the cluster lock disk reference in the cluster ascii file to disks in new array.
1) Can the cluster lock disk in 2 node cluster be the last PV in the shared VG ?



2) I got raw devices in VG ops (see below). The old array had 36684 X 4 =aprrox 145gb space assigned to ops vol .group/
Can I just allocate 14574 (alloc PE) X 4 = approx 58gb since the vgdisplay shows only that much being allocated and then mirror the lv from old to new array. I dont want to give VG more space than it needs.
--- Volume groups ---
VG Name /dev/ops
VG Write Access read/write
VG Status available, shared, server
Max LV 255
Cur LV 81
Open LV 81
Max PV 13
Cur PV 1
Act PV 1
Max PE per PV 36684
VGDA 2
PE Size (Mbytes) 4
Total PE 36678
Alloc PE 14574
Free PE 22104
Total PVG 0
Total Spare PVs 0
Total Spare PVs in use 0


If anyone has steps to migrate, it would be nice too so I can cross reference my work.

Since it is ops I need to shut down cluster, modify cluster lock disk to new device in newarray, add new disks from array and mirror old to new. Remove old array luns. Reactive cluster in shared mode . Start app.
Thanks

[Steven E. Protter]

Shalom,

1) Yes. Any disk accessible to all nodes is acceptable as a cluster lock disk.
After changes.
cmquerycl
cmcheckconf
cmapplyconf

No downtime required.

2
I might think that mirror/ux might be useful in replicating logical volumes if both shared disk arrays are visible at the same time.

You may be able to use vgextend to extend the shared volume group, but creating a new one might be easier.

SEP

[Sammy_2]

Steve,
Thanks for the reply. I have always needed downtime (shutdown cluster) to make any changes to Vg in shared mode (we have 2 node RAC). Need to activate the VG in exclusive mode from one node and then make changes (even extending a LV). Let me know if you think I am doing anything wrong. (See the instruction below that I use)

I will lvextend the LVs once I add the luns from new array to vg. I am going to vgextend because if I create another VG , then I can't mirror across VGs.


ops is shared vg.

ON SERVER and CLIENT NODES
==============================
# vgchange -a n /dev/ops (deactivate the vol. group)
# vgchange -c n /dev/ops (make cluster unaware)

ON SERVER NODE
====================
# vgchange -a y /dev/ops (Activate ops vg before extending it)
# lvextend -L 2000 /dev/ops/dtictbl04 (extend the LV)
# vgchange -a n /dev/ops (deactivate the vol. group before makeing cluste aware and shareable vg)
# vgchange -c y -S y /dev/ops (mark /dev/ops vg as shareable vol. group)

[Hasan Atasoy]

hi sammy ;

if you installed snor patch you dont have to close the voluem group on both nodes. you just need one node to be closed. please look at the document ,

docs.hp.com/en/7389/LVM_SNOR_whitepaper.pdf

Hasan.