We're having reliability issues with just one VM among the 25 we're hosting on this host server.
About this VM:
OS: Windows Server 2008 R2 SP1
Software: Office 2010 and a line-of-business application called AVImark. Actually, the users start the executable with a shortcut that points to the EXE on another VM. Other VMs start the same EXE residing on the same server but these VMs don't have these symptoms.
After a few weeks, users will report the following and in this order:
- an inability to logon via RDS
- no Desktop icons (folder redirection fails)
- eventually existing RDS sessions become unresponsive.
Other symptoms from the console session:
- numerous service fail:
- ttkEvents: "Insufficient system resources exist to complete the requested service"
- WinLogon: "The Windows logon process has terminated unexpectedly"
- Folder Redirection: "Failed to apply policy and redirect folder..."
- Group Policy Drive Maps: "not enough storage is available to process this command"
- everything slows down. What slows down? Everything. Clicking anything might result in a 10s, 20s, 60s or longer delay before getting a response. Eventually, no input gets a response and the VM has to be shutdown which it will do gracefully if the shutdown is initiated by Hyper-V Manager. However, I found that if I catch the VM in a semi-usable state (before it completely locks up), I can resuscitate the VM by disabling then enabling the NIC from within the VM. Things aren't perfect - there's still some delay in clicking things, Resource Manager shows no data, etc. - but it'll get them through the day until I can reboot the VM at night.