Extending a volumegroup with mirrored disks

[P. Overduin]

Hi All,

I have a question regarding extending volumegroups and mirroring.

I am working with a MS/SG cluster (2 nodes) which are using 2 identical EVA3000 SAN's. All volumegroups are (Mirror/UX) mirrored over those 2 SAN's.

One of the volumegroups is running out of space and I created an extra LUN on both SAN's. The plan is to use vgextend and let the disks join in the existing VG.

Initially I configured the volumegroup using "lvcreate –m 1 –L /dev/vgname" thus creating a mirror over the 2 virtual disks.

So far I have created 2 physical volumes (pvcreate /dev/rdisk/c*t*d*) which I would like to mirror and extend into the existing volumegroup. I am getting a bit lost now in the PV LV VG and so on forest and don't want to *&^ this thing up the last week of the year.

My questions are:
Should I use vgextend to join those 2 new PV's into the volumegroup?
How do I get those 2 new PV's to mirror in the way the 2 existing disks do?
Will there be dataloss (do I have to schedule a restore) and/or must I take the VG offline?
Does vgexport and vgimport on the secondary node take care of the new configuration?

I hope some off you can help me out.

Kind regards,

Pete

[Ravi_8]

hi pete

break the mirror, add the new disk to vg and recreate the miror

[Stanimir]

I suppose it is enough to extend your existing VG using:

1. When you create LUN on EVA3000 you can
assign RAID1-mode to those
ph.disks,which are included on this
virtual LUN. /The task of SAN-admin/
It is enough to mirroring. No action
needs from the side of OS.

2.Find this new LUN from OS, using:
#ioscan -fnC disk
#insf -e

3.Extend existing VG using new dev.files:
#pvcreate /dev/rdsk/c20t4d0
#vgextend vg01 /dev/dsk/c20t4d0
#lvextend ...
#extendfs ... and so on


Regards,Stan

[Stanimir]

In addition if you need mirroring on OS anyway, you can do:

#lvextend -m 1 /dev/vg01/lvol1 ...

[Hoang Chi Cong_1]

Hi.
This is a thread with the same case.
You can see it:
http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=774044

Regard,
HoangChiCong

[P. Overduin]

Thanks for the rapid reply's !

I have to say it helps me in to the right direction. I should explain the use of 2 EVA's here. They are used to create a redundant environment and placed on different locations. That is why I used (v)raid 5 on the EVA and Mirror/UX between both EVA's for each Unix-used LUN.

I hope there is somebody that can help me out with the other open questions.

Peter

[Sunil Sharma_1]

Answer of your second question is Yes, vgexport and vgimport with -s switch will take care of new configuration in secondery node.


Sunil