SUN Volume manager migration

never tried it but here it is:

http://docs.huihoo.com/opensolaris/solaris-volume-manager-administration-guide/html/ch25s09.html

i would clone/snap your volumes before you try this, just in case you need roll back

thanks a lot

Hi, TomaHawk,
Sun SVM is one of the less flexible volume managers I work with, compared to Veritas, AIX LVM, or HP LVM. But what you're trying attempting is possible.

First, my assumptions.
1) You want to move a set of luns from one server to another server
2) Both servers are of identical OS levels and patches
3) Both servers see the luns simultaneously during this transition time

Here's what I would do.

Prep - can be done online
1) on original server: metastat -p
2) on target server: metastat -p >> /etc/lvm/md.tab
3) copy the output of "metastat -p" from the original server and append it to the md.tab file on the new server. Just copy and paste. This sounds strange but the output of metastat -p will be used as a roadmap for SVM to import the luns later on the target node.
4) Edit md.tab on the target server and change only the device file references. Because you copied it from the orig server, the device file may (probably will) be different on the target server, even for the same set of luns. For example, a LUN that is called emcpower0c on the original server may show up as emcpower1c on the target server. As a result, edit the md.tab on the target server to replace emcpower0c with emcpower1c. Use "powermt display dev=all" on both nodes to help you figure out this relationship.
5) Backup all your data

Outage
1) Power down the original system
2) On the target system: metainit -a
This will build the SVM objects on the target system, using your modified /etc/lvm/md.tab as a guide.
3) On the target system: metastat -p
This should show you your SVM metadevices, soft partitions, and any disksets you might use.
4) Mount your filesystems on the target system.
5) While the original system is still down, remove access to the storage by unmasking or unzoning.