VMware Horizon Community
bhamm
Enthusiast
Enthusiast
Jump to solution

Moving Linked Clones from Local Storage

Hi,

My department is in the process of upgrading to vSphere 5 and View 5.  We have several linked clone pools on local storage.  I want to move the pools to another host in order to upgrade the host's hypervisor, as well as to (eventually) upgrade the pool's agent, hardware, and vmware tools.  From what I remember, in order to move the pool, I'd have to recompose the pool onto the other host/datastore, but for some reason when I did a test run, the pool won't move.

Here's a rundown of what I'm trying to do:

  • I have a linked clone pool with persistant disks (dedicated assignment) on Host A.  Host A has local datastores A1 and A2.  The replica is on A1, and the deltas are on A2. The persistent disk is on shared storage, which I'll call S1.
  • I want to move the pool to Host B, which has a similar structure.  It has local datastores B1 and B2.  The P disks would stay on S1, with the replica & deltas being on B1 and B2, respectively.
  • Host A & B are part of the same cluster. S1 is visilble to both hosts.

For my test run, i did the following:

  • I created a test pool with 1 vm matching the description of the pool above.
  • I logged in to the VM and made some changes so I could confirm the p-disk had been restored after recomposing.
  • I created a new snapshot of the parent VM (w/ minor changes to the system so I could ID it).
  • I logged onto the View Administrator, selected the test pool, went to Edit->Vcenter Settings, and selected datastores B1 & B2.  I deselected A1 & A2.
  • I saved the settings, and then recomposed the pool with the new snapshot of the parent VM.

When I did that, the recompose operation went off in vsphere.  A new replica was made, and the old one was deleted.  I was able to log into the VM again.

However, when I checked the settings, the VM is still running on Host A, with datastores A1 & A2.  I checked the settings in the View Admin console, and confirmed that B1 & B2 were selected for the pool.  A! & A2 were not.

I tried it again with a different VM, not just a different snapshot.  I got the same results.  The recompose operation happens, but the pool stays on the original host/datastores.

Any idea what I'm doing wrong?  I've looked through the community pages, and this is generally what i've seen as far as the recommended way to move the clones.  One post said to select the new datastores in the recompose wizard, but I don't see that option when I recompose my pool.  I assume this is because that post was for an older version of view.

My environment is:

Hosts - ESX4.1

View 5 Connection Server

Vcenter 5.0

VMs w/ agent 4.6 & Hardware v7

Thanks in advance for any help.

0 Kudos
1 Solution

Accepted Solutions
kgsivan
VMware Employee
VMware Employee
Jump to solution

Recompose across ESX's Local datastore is not recommended and not supported.

But there is an easier way since your Persistent disks are on shared Storage

Here is the way how to do.

Let us call your original pool as pool 1. Pool-1 Replical will be in Local datastore 1, delta on local datastore 2, and persistentd disks on Shared datastore 1

Create a new Pool in host 2 as of the same structure in host 1 (same pool size too) Let us call this sa pool 2. Create this pool in on demand provisioning

Delete pool 1 archiving its persistent disks

go to persistent disks page, go to detached page-  edit each persistent disk, change the 'Last pool' to pool-2

now select all persistent disks and click on re-create

all the VMs will be recreated in pool -2

You can also try, delete the pool-1 first and create a pool in host 2 in the same name 'pool-1' and recreate persistent disks withour editing it. ,- this you need to really try out. but the above solution will work perfectly

Thanks

Skg

View solution in original post

0 Kudos
4 Replies
bhamm
Enthusiast
Enthusiast
Jump to solution

I should add: I've tried the rebalance feature as well.  I change the datastores in the same way as above, and run the rebalance command in the View Admin console.  vSphere shows the error "Unable to access the virtual machine configuration: Unable to access file [b2]", presumably because the replica is not visible from the new host. This occurs even if i select a store on Host B for the replica image.

0 Kudos
kgsivan
VMware Employee
VMware Employee
Jump to solution

Recompose across ESX's Local datastore is not recommended and not supported.

But there is an easier way since your Persistent disks are on shared Storage

Here is the way how to do.

Let us call your original pool as pool 1. Pool-1 Replical will be in Local datastore 1, delta on local datastore 2, and persistentd disks on Shared datastore 1

Create a new Pool in host 2 as of the same structure in host 1 (same pool size too) Let us call this sa pool 2. Create this pool in on demand provisioning

Delete pool 1 archiving its persistent disks

go to persistent disks page, go to detached page-  edit each persistent disk, change the 'Last pool' to pool-2

now select all persistent disks and click on re-create

all the VMs will be recreated in pool -2

You can also try, delete the pool-1 first and create a pool in host 2 in the same name 'pool-1' and recreate persistent disks withour editing it. ,- this you need to really try out. but the above solution will work perfectly

Thanks

Skg

0 Kudos
amandasmith
Enthusiast
Enthusiast
Jump to solution

Hi,

Check the following information http://communities.vmware.com/message/1872966

Hope this information helps you. http://spaceship.sytes.net

bhamm
Enthusiast
Enthusiast
Jump to solution

Thanks guys for taking the time to reply.

I'm  surprised to hear that recomposing to a different local datastore isn't  supported.  Since the master VM is located on shared storage, it's  reachable by the 2nd host.  A new replica would be made for any update,  so it could be put on the new host.  What's the technical difficulty in  recreating the the pool on a different datastore via recompose?

I've  read that there was a change in View 4.5; originally, the VMs were  deleted entirely for a recompose, but now some of the config info (e.g.  AD computer account info)is maintained on an internal data disk.  Is it  info such as this that makes this difficult?

I'd rather avoid having to remove/reattach each disk on a large pools, as well as come up with a new name scheme for each pool when they're moved, if possible.

0 Kudos