migrate from storage array to hardware raid

[CSG Office]

Hello everyone,

I have a T520 running HPUX 10.20. Currently, the volume group containing all of the system partitions is located on non hardware raid storage. I want to move these partitions to a hardware raid. This is what I would like to do:

1. Add physical devices to the current volume group.
2. Use pvmove to move the data off the non raid storage.
3. Use pvreduce to remove the non raid devices from the volume group.
4. Reboot.
5. Reset the PRIMPATH to the new raid device's hardware address.

Am I missing anything? Would this work?

[Simon Hargrave]

Yes this will work however a couple of considerations: -

- You may have problems depending on how your volume group was configured, in terms of maxpv etc (ie if you can't add enough extra PVs to your volume group).

- If you have mirrordisk/ux I would choose to lvextend each logical volume then lvreduce the non-raid portion rather than using pvmove. Whilst pvmove will work I'd always be happier having 2 lots of my data than risk a failure during pvmove.

- I'm sure it was a mistype, but you'll need lvreduce not pvreduce to remove the non-raid.

[Mel Burslan · Accepted solution]

At step 2, instead of pvmove, I personally prefer adding new drives as a mirror or second mirror. I found out that the mirror replicating being faster than pvmove.

between steps 3 & 4, after the mirroring completes, you will need to make these volumes bootable by executing few commands as follows:

rootdisk_raw=/dev/rdsk/cXtXdX
mkboot -a "hpux -lq (;0)/stand/vmunix" ${rootdisk_raw}
mkboot -l ${rootdisk_raw}
cd /usr/sbin/diag/lif
mkboot -b updatediaglif -p ISL -p AUTO -p HPUX -p LABEL ${rootdisk_raw}
lvlnboot -r /dev/vg00/lvol3 # root fs /
lvlnboot -s /dev/vg00/lvol2 #swap
lvlnboot -d /dev/vg00/lvol2 #swap/dump
lvlnboot -b /dev/vg00/lvol1
lvlnboot -R
lvlnboot -v


hope this helps

[Geoff Wild]

Looks good - I don't believe you need step 4 and 5 though...the primary path will be set when you do the lvreduce, and the vgreduce.

Like Simon - I would use mirroring if available.

Rgds...Geoff

[Steven E. Protter]

Your approach will probably work.

I would do it differently.

1. Start with a good, cold backup of all data.
2. Use lvextend -m 1 to make mirror copies on the new storage, which has been added to the volume group and use lvsplit to break them apart later.

Note: Your volume group unless created properly might not be able to handle the new storage and the old at the same time. This is typical over 50 GB when special options -p command is not used to create the volume group and limit the number of physical volumes.

If this is the case you may have to use fbackup to write a backup and restore, or create a new volume group and use the cp command to copy the data.

In either event, you'll need to schedule downtime to handle that.

SEP

[CSG Office]

The Mirroring package is not currently installed or I would go that route. I don't really want to install it since the new storage is hardware raid anyway. Also, downtime must be reduced to as little as possible because this is a 24/7 server.

So, since there seems to be support for the pvmove method working, I will try that.

Thanks everyone.