Hello,
I'm using Horizon with View 5.3.1 in vSphere 5.5.
We are in a test phase for a Win7 linked clone pool. After recompositioning a guest it looses its OS. It stays in a boot loop telling me that the OS can't be found.
Any advice?
AWo
http://kb.vmware.com/kb/2001345
I saw this KB and thought of you. Not sure it helps but grasping at straws at this point.
So no issues on the parent machine but when you deploy to the pool you get that No OS found? Is the parent machine fairly standard (accepted all defaults) or is there some custom hardware in place, like a pvscsi controller?
Your're right Sir. Fairly standard hardware on the parent, nothing special. And I get that no OS found during the recomposition of a linked clone, which already worked. The initial deployment works, then I'll test recompositioning a guest and that's where the problem comes in.
AWo
So a recompose to a different snapshot other than the one originally deployed? Have you by chance created a third snapshot and tried a recompose based on that? Just to rule out something weird with the second one.
I recompose to a second snapshot of the original parent.
The guest is left in the status "customizing". I tried to recompose it to a third snapshot and that leads to a switched off guest and a customizing timeout failure in the View Desktop list. I tried to repeat that action, but the guest stays switched off.
AWo
BTW, I did the test with and without persistent disks (tested with two different pools) and with and without host caching enabled.
AWo
So it was in a stuck status of customizing, you powered it on and it then booted with the can't find OS message? Sometimes the stuck customizing status can be the result of network issues. Does everything check out there, including DHCP Scopes?
There are enough DHCP addresses free and as the MAC address didn't change (or does it change?) it should get its old one back.
Would a listing of the files of the guest help you? I could provide a list of a guest provisioned freshly and working and what is the case after recomposing.
AWo
Any extra information is great. On the recompose to the second snapshot do you see the replica being created?
Healthy guest:
No OS guest:
AWo
What about the replica for the second snapshot? Is it being created when the recompose is happening and is the host running that VM able to access the datastore that is storing the replica?
mittim12 schrieb:
What about the replica for the second snapshot? Is it being created when the recompose is happening and is the host running that VM able to access the datastore that is storing the replica?
How to find out that?
Yes, the guests are stored on the same datastore and run on the same hypervisor. All datastores are accessible by all hypervisors.
AWo
The listing of the parent:
AWo
I need to clarify something. Is this just a single guest having the issue or all guest having the issue when recomposing to the second snapshot?
If it's the latter then I would be curious if the replica is being created or failing when the recompose is happening. On the first deployment of a snapshot a corresponding replica is created. All the clones based on this snapshot will communicate with this replica via read IO. The replica VM is always created with the prefix of "replica" as seen in the screen shot. As far as I know if the clone can't access the replica then it will fail to boot although I thought it simply wouldn't power on.
I created a pool with max. 10 guests. Up to four were created during testing. All four guests are facing the problem, so the latter applies.
There is a directory called "replica-........" with these files in it. Is that what you were looking for?
I'll checked the time when the snapshots of the parent were created (is that relevant?). The file list shows 12:50, the snapshots were cretated 12:28, 14:31 and 14:51.
I have three snapshots from the parent, now. All from the first snapshot (v1). And three guests in the pool. Does that correspond to what you see in the listing?
I go and change the pool to snapshot v2 and create another guest.
BTW, a new guest based on a newer snapshots of the parent (the snapshot I want to recompose the other guests to) is working. Until I recompose it for the first time.
AWo
After switching the whole pool to the second snapshot and creating a new guest, a new replica directory was created.
AWo
Ok, that is what I would expect to see. Did that guest come online correctly?
In principle yes. That one first runs into the Microsoft Windows Repair process. After canceling that the guest bootes normally and run into the VMWare image process. After that it was fine.
AWo
But the problem still exists. Aftre the first recompositiuon no OS can be found.
AWo
AWo wrote:
In principle yes. That one first runs into the Microsoft Windows Repair process. After canceling that the guest bootes normally and run into the VMWare image process. After that it was fine.
AWo
I'll have to be honest I'm not sure what is going on. It might be best to start from scratch with a different image. Does the note above about running the VMware image process imply you are using sysprep?
