A week ago, I fired up a (3) Host Cluster with dedicated VDM and VC for our operation in Bangalore, India. All has gone well until my Non-Pers Pool of 46 VM's get's it's 20th User logged in.
The next users to try to logon to the Pool using the Web Access version get logged on and see the VM's or Pools available to them, but if they try to logon to a pool that already has 20 Users logged into it, they will see the Virtual Desktop Window try to fire off and then it will disappear. Users connecting with the VDM Client will get an error:
There was an unexpected RDP disconnection. Microsoft RDP error code = 3078
The User can log on to any other Pool or VM they wish. This only happens when logging into a Pool that already has 20 concurrent sessions on it.
We have checked our Virtual Switches and all are set to the 56 default. We're still scratching our heads. VMware Tech Guy Bill hasn't called us back in 3 days as I can only assume he's scratching his head as well.
Any thoughts out there? I thank you in advance for any advice on tracking this issue down....
There have been several posts here about this - clearing a couple of registry values on the VMs should fix it, if it's happening for all VMs in a pool, you should clear them from the template. Recent post by Stevester: http://communities.vmware.com/thread/154460
Thanks for your reply.
This is not happening to all VM's in the pool. I stated it would happen after 20 VM's were connected to by Users. Another User trying to connect (for lack of a better term, the 21st User) would get this error and all subsequent Users trying to connect to the pool would get the same error. If there were only 19 or less Users connected to the pool and working along just fine and another User connected, they wouldn't get this error. Only after 20 Users are connected do new Users get this error.
This is not entirely surprising, as your users are probably only attempting to connect to set of affected VMs after a certain point - they are picked based on availability and response time to the broker, so with all factors being the same you will see some VMs reguarly connected to earlier than others in a pool. I would recommend clearing the key on all VMs that don't have a session on them and seeing whether that resolves the issue, it should be harmless to do this for working VMs as it's regenerated.
It's not surprising that end users would end up attempting to connect to the same VM at roughly the same saturation point. If it's exactly 20 every time though, that would warrent further investigation into your pool setup - are all fifty provisioned and powered on, for example? What I was trying to say is that it's very possible that a couple of VMs in this state could appear to affect the whole pool as they are the most likely to respond at a certain point.
This question has been answered by Jose Perez at VM Tech Support.
The issue was the Pool Settings. Making a pool in 2.1 without using the Advanced Pool Settings options was what stalled any connections over 20.
Once we enabled Advanced Settings, the rest of the VM's became available.
Jose states he was told by the Product Engr'g Dept that this was the "fix".