Need to "Back Out" installed drives

[Paul J. Ledbetter, II]

I recently installed two ST373405FC drives in my VA 7400. The Array is set up in a RAID 10 configuration, with the drives in Redundancy Groups 1 & 2 respectively. I was unaware that auto-include was enabled, andf they are now part of my current RAID configuration. I need to back these drives out and start over. How can I do that?

[Patrick Wallek]

You really don't have a choice but to have the drives be part of the RGs in the VA7400.

If they are not included in the RGs then you will not be able to use them in the array. You can NOT treat them as normal JBOD disks.

What exactly were you hoping to accomplish with these drives?

[Jayasuntar]

Hi,

No choice of removing the disk from RG, once it is added.

Jay

[Paul J. Ledbetter, II]

My original intention was to create a separate RAID-1 LUN across these two new 73 GB drives. I've since done some reading and it appears that I'm stuck with my RAID 1+0. But I still want to create a single LUN (actually VG) across these two drives so I can have a place for my programmers to dump data to when they defrag their databases. I originally had 6 36 GB drives and 2 18 GB drives comprising 2 separate VGs on my array. All I want to do is to create a third LUN on the array and then define my VG and be on my way. I need to know how to get back to my original configuration before I put the drives in. I will then disable the auto-include and (hopefully) create my third LUN and be done.

[Uwe Zessin]

The VA7400 cannot do a traditional RAID-1 with 2 disk drives. By default all disks on even numbered slots are assigned to one RG - the remaining disks are assigned to the other RG.

[Patrick Wallek]

What you desire is not possible with the VA7400. The only thing you can control when creating a LUN is the RG that it is in. The LUN will then be created across ALL disks in the RG you choose.

With the VA7400 the disks in the odd-numbered slots are all in RG 1 and the disks in the even numbered slots are in RG 2.

So no matter how you add the drives, in order for them to be usable in the VA they must be included, whether automatically or manually doesn't matter, and they will then be part of the RG appropriate for the slot they are plugged into.

[Paul J. Ledbetter, II]

I'm OK now with them being in the same RG. I just want to create a new LUN and hence a new VG with my drives. Can I stick drives in one RG and not have the same type of drive in the other, corresponding RG? Can I have, for example, drives in slots 9 & 11 (RG 1) without having the same kind of drives in 10 & 12? Am I limited to 2 LUNs on the VA7400? What happens if I just remove them from 9 & 10 now? Will I have database corruption?

[Patrick Wallek]

As far as I know, yes you could have drives in slots 9 & 11, without having drives in slots 10 & 12.

No you are NOT limited to only 2 LUNs, far from it. But you are limited in how you create the LUNs. A LUN can be created in either RG1 or RG2. Not both. An you can not specify a granularity smaller than the RG.

[Paul J. Ledbetter, II]

It seems my vendor who originally installed my VA7400 did something "sneaky" and got around normal VA7400 behavior when they initially set me up.

Despite my original instructions, they set me up with 3 x 36 GB and 1 x 18 GB drives per RG. From discussions with other HP engineers, I should not have gotten the (3*33.37+16.68)/2 = 58.39GB that would normally have been expected from a RAID 1+0 configuratiion. The mismatch of the one 18 GB drive should have given me something less. Apparently my vendor was able to "force" the VA to give me all expected space. Well they didn't tell me.

So when I added my 73 GB drives on each RG, I expected similar behavior, and of course I was disappointed. I only got an additional 16 GB instead of the expected 33. Removing the 18 GB and replacing it with a 36 GB in each RG helped get me the additional disk space I was missing.

My original request was just asking for the feasibility of immediately removing the 73 GB drives to get back to my original configuration so I can re-analyze me situation.

Thanks for all of your attempts to try to understand what I needed.

[Paul J. Ledbetter, II]

I have "solved" this problem.