Implementing internal raid 1+0 after O/S has been applied

[Doug Short]

I have a customer that has some flavor of HP blade server, with 3 internal disks. He is new to the sys admin role so when he built the O/S on the internal disk, he did not know that he could create a raid group with the other disks he has installed. After realizing that he also did not purchase MirrorUX, he is now coming to the realization that his new Blade server has no fallback should something happen to the internal drive that he's up on. This box is not in production yet, so there is time to resolve his issue. My question is, can I put the internal disks into a raid 1+0 group with the 3rd disk being a hot spare, without being data destructive and having to rebuild his O/S? Or should I push the customer towards purchasing MirrorUX to give him some level of redundancy? I would rather see this server in a raid grouping as my personal opinion is that it is a better fallback, but given that a new production date has just been applied, I'm looking for a quick solution. Any assistance is greatly appreciated.

[Torsten.]

If you have 3 disks as you say, I would simply create a hardware mirrored pair on the empty disks and use DRD to clone the existing boot drive. After this you can boot from the hardware mirror and assign the remaining disk as hot spare (if you really feel you need this).


Unfortunately creating the mirrored pair is destructive ... so doing the above will only need 2 boots (1 for accessing the adapter BIOS, 1 for the final boot) and that way you keep your data.

[Viveki]

I believe the blade system is bl870C as you say 3 internal drives. This system comes with an embedded LSI-1068 SAS RAID controller, supporting RAID 1 for HP-UX and Linux.

Please restart the system and press CTRL-C when indicated to begin RAID configuration and enter the LSI Logic MPT SetupUtility.

But I think you need to start from scratch again.

[Torsten.]

Regarding the CTRL-C (which is wrong) see

http://docs.hp.com/en/5900-0112/ch03.html

and no need to start from scratch, see DRD instead:

http://docs.hp.com/en/oshpux11iv3.html#Dynamic%20Root%20Disk

[Doug Short]

I sincerely appreciate the fast responses. Torsten, I like your idea of DRD better and will propose that to my customer. Ultimately it's his decision on how he wants to manage this server and how he moves forward. DRD is new to me, and I was about to ask what it was when you provided the link to the documentation, as when I build machines from scratch, I normally create the raid group prior to laying down the O/S from the beginning. Which of course is the other option here, starting again from scratch.

Thanks again for the prompt responses, I will close this thread and assign points.

[Doug Short]

Likely doing DRD, but will give the customer the option of building from scratch as well.

[Torsten.]

You are welcome.

Do you know about this already?

http://forums11.itrc.hp.com/service/forums/helptips.do?#33