A hypervisor is like a bank. It is offering inexistent CPU cycles and HDD I/Os up to 150% of real capacity. In normal situation VMs consumes about 20% of declared (maximum) RAM, CPU and disk IO, and business is running perfectly … until … Until everyone will come to bank and ask "I want my money back". Good example, short power outage or unexpected hypervisor reboot. Once the hypervisor will start it will try to power on all VM guests, and every of them will want "his money back". As a result we have temporary IO storm and after 10 – 15 mins everything backs to normal. VM guests works, however many of them timeouted some services, for example MSSQLSERVER J. Let's increase timeout delay on our VM guests.
Windows Registry Editor Version 5.00
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control]
"ServicesPipeTimeout"=dword:00180000
Since now our services have 3 mins to start … and to stop when you shutting down the VM J