changing the name of a shared datastore

[brucecmc]

running esx 3.0.2, VC2, all patches applied.

I have an HA/DRS cluster configured. I have CX400 Clariion Backend storage. I present ESX1's storage to ESX2 and vise versa.

In VC, on ESX1, the storage displays as "snap-0000000-04-ESX2_DS, and vise versa for ESX2.

I have ran across this previously and it was just a matter of renaming, on ESX1, the correct datastore name (ie...ESX2_DS).

However, in this case, I'm getting an error that there is a datastore with this name already and I'm not sure what to do about correcting the name within VC. It doesnt appear to be harmful, but obviously, I'd like my datastores on both ESX's to be something I recognize easily.

Any assistance?

thanks

Bruce

[RParker]

If you click on Inventory -> datastores you can see ALL of your datastore names. Are you sure there isn't another one someplace?

That's probably why the error.

[ewannema]

I am not a SAN expert, but I think that you should look and make sure that you are presenting the LUNs the same way to both hosts. Make sure the LUN numbers match. I believe that the SNAP name shows up when the computed signature does not match what is written to the disk.

I am guessing that you set /LVM/DisallowSnapshotLun to 0 and without this setting you would not be able to see the disks at all.

[brucecmc]

RP,

Yes, thats the very case...

On ESX1, I created a datastore called ESX1_DS. I then presented the LUN also to ESX2. On ESX2, it displays as snap-000000-04-ESX1_DS. when I try to change it, it definately sees the datastore as being present...

It's confusing me a bit, because, I have multiple datastores presented to these 2 ESX hosts, some of which I had to remove this "snap" caveat. But, these 2 are just giving me a run around...

the LVM allow resignaturing is enabled (tried both enabled and disabled 0 and 1) as well as the allow/disallow snap shots...

I'm just missing a little piece here somewhere...

bruc e

[ewannema]

If you are not cloning disks then you should not need either of those settings. Make sure that the LUNs are being presented with the same #s for each host. If they are not they will be seen as snapshots. This guide is for Symm storage, but it still contains a lot of good information: .

If you continue to have problems, I would focus on getting everyting to look correct on one of the hosts and then move to the second one.

[brucecmc]

that could very well be the problem...

Understand the issues with LUN numbers in a symm environment. However, this is an older CX, non-production, doing some testing and such and only have 1 Storage Processor. So, all LUN's are owned by this single storage processor and I would think therefore not able to duplicate the LUN numbers...

Hmmm...might be the issue all together...

I'll keep digging...thanks for the thoughts...

bruce

[bolsen]

On the EMC the LUNs have to have the same ID in each host group, storage processors have nothing to do with it. You can specify the ID when you intially add the LUN. (Scroll to the right.) If you need to change the LUN ID you'll have to remove the LUN, readd it and then change it.

If your LUN IDs are the same, it could be a resignature problem. Check out the SAN CFG guide and search for resignaturing.

[brucecmc]

thanks for the reply...they definately are NOT the same LUN numbers...I will give that a try...I dont know that I'll be able to assign the same LUN number to 2 different devices...but, will give it a shot...

will post here results...

bruce

[brucecmc]

bolsen,

after thinking on that for a minute...this isnt 2 different LUN's that I'm talking about...from the SP when presenting the LUN to the ESX Host(s) it is the same LUN... ESX1 see's the LUN and ESX2 see's it as well. I misread your post...but, i'm talking about the same LUN...

And I agree, I thought this was just a resignaturing issue, but i've tried changing the LVM settings to allow/disallow resignaturing...

bruce

[ewannema]

Resignaturing a disk is generally not a good thing since that signature is used to determine the paths to the VMX and potentially vmdk files. You should not need to enable resignaturing to get multiple hosts in a cluster to see the same storage.

Say you have a 20GB LUN you are presenting. Are the LUN #s the same for this on both hosts? In VIC you can go to Configuration | Storage (SCSI, SAN, and NFS) for the hosts to verify. Say Host1 vmbhba1:2:5:1 & Host2 vmhba2:2:5:1. If not, then you have a problem.

[brucecmc]

Based on how I read this, the LUN numbers are the 3rd number in the set, correct..In that case, No, they are not the same LUN numbers on both hosts...

one has it listed as vmhba1:0:0:1 the other is vmhba1:0:2:1

bruce

[bolsen]

Looks like they are different lun ids.

Login to the EMC console and open the LUNs under the hosts. If you scroll right it should show the LUN id. If they are different (even though they are assigned to different hosts, the ID has to be the same) then you'll have to remove the LUN from one of the hosts and re-add it as the same LUN ID as the other.

[brucecmc]

Well folks, I'm gonna have to wrestle this one later...I've got some other pressing issues that I must attend to...I dont suppose my problem is really resolved...If anybody has any further thoughts, please feel free to throw them out here...

I'm not exactly sure how I'm suppose to get the LUN numbers to match on both ESX hosts as the LUN number that is being assigned, does not coincide with the LUN numbers from the storage array...

thanks for the thoughts and help so far.

Bruce

[stvkpln]

The issue is basically this: you have two hosts that can see the LUN's, but the LUN's are not being presented as the same LUN ID on each of the ESX hosts. The easiest way to resolve the issue is to do the following:

1. Remove any relevant storage groups to these two systems immediately.

2. Create a new storage group on the CX and call it whatever makes you happy

3. Take all the LUN's you want presented to these two hosts and set them in this storage group

4. Take the two hosts and put them in this storage group.

5. Reboot both hosts

6. Check the storage, all should be ok now.

As a side note, I'd highly recommend disabling both the Resignaturing Option and the DisallowSnapshot option. By default, you should never need them if your storage is being properly presented to your hosts.