Hi all,
I've run into an interesting problem today while working on getting a Windows 7 pool working. I was having an issue where sometimes desktops would have a black screen upon connecting with the view client, so I tried the solution at http://thejeremykeen.com/2010/03/16/blank-screen-on-windows-7-vmware-view-desktop-using-pcoip/. When I took a new snapshot and recomposed the pool, I started to have problems with not being able to log in. When I try to log in a test user, I get the message that the user profile service encountered an error, and couldn't load the profile. I double-checked, and it isn't happening on the parent VM. So I tried a few things, one of which was to try to log into the clone via the console with my own domain account. It logged me in, although it did load the default profile instead of my own for some reason. I also noticed that the VMWare Tools service is crashing, and that the persistent and non-persistent disks were showing up, but I couldn't open them like normal. At this point I figured I had somehow screwed something up horribly, and recomposed the pool back to the snapshot that it was using before I started messing with all this.
Here's where it gets stupid. Even with this snapshot I know for a fact is good, the clones in the pool are giving the error about not being able to load the profile when I log in. What could cause this? I've deleted and re-created all the machines in the pool before this point, so I don't think it's that. But I'm mystified as to what could possibly cause this issue.
WHen creating the master image, did you login with the same account? I never login with a normal account, if I do need to login to the master image, I use a test domain account which will never login to a LInked CLone machine. The profiles can clash
I did not, except when I started having the problem and tested to verify that the test account I was using worked on the parent VM. I've tried several accounts anyway, and all of them have issues. Some are able to log in, but the extra disks aren't working right, the profile isn't getting created (you get the message about how Windows logged you on with the default profile), and the VMWare Tools service crashes when you log in.
Hmmm very strange. I have no idea but I would run this test:
Convert the image to a template and deploy to a full virtual machine - does this have the same behaviour? If not, there is quickprep and linked clones that could be the culprit. Does the same occur on a different datastore? Different host?
Anything in the event logs? Can you change the pool type to not use a disposable disk, or a persistent disk - does the same happen then? Then introduce each disk one at a time to see if it occurs then
So, after some messing around, out of sheer desperation I tried to apply the next-oldest snapshot to the pool, and it worked. So the host and datastore is all good, but somehow this (previously good) snapshot got corrupted. Very strange, as I thought that once a snapshot was taken it couldn't be modified.
Long story short it's good now, I re-did the changes that I had made to the base VM before and it's still working. I have no idea what could cause the problems I was seeing (at one point I even checked and saw that the logical disk manager in Windows could see the extra disks, but they still didn't show up properly in Explorer), but perhaps it's not worth worrying about if I have managed to sidestep the issue.
