Windows EFI VMs fails to start if it has assigned RAM more than 64256 MB (63.75 GB).
The guest OS reboots in the cycle, it is not possible to manage the VM via VNC or using PMC.
In the log file of VM, the following log records can be found every time the guest OS tries to start:
10-08 18:08:38.529 F /vm:739012:816091/ Sending question = PRL_ERR_CPU_RESTART SystemReset by triple-fault 10-08 18:08:38.530 F /vm:739012:816091/ VM state(VmStateRunning): enqueued 'VmLocalCmdHardwareReset'(20009) command 10-08 18:08:38.530 F /vm:739012:816091/ VCPU0 state(VcpuStateRunning): enqueued 'VcpuCmdTerminate'(6) command
This issue has been confirmed and the internal request #PSBM-26752 submitted to have this issue fixed in future updates.
There are two possible workarounds.
Disable EFI boot for this VM. The installation of Windows should be repaired for proper boot.
~# prlctl set VMName --efi-boot no
Extract the attached archive with EFI boot image on the server, and restart VM via stop/start.
~# wget http://kb.plesk.com/Attachments/kcs-38348/lin-efi64.tar.gz ~# tar zxf lin-efi64.tar.gz -C /usr/lib64/parallels-server/ ~# prlctl stop VMName ~# prlctl start VMName
Note: It is necessary to remove this file once the official fix is released: no special check is done during updates for presence of this file. Watch the release notes for the fix of #PSBM-26752.
~# rm /usr/lib64/parallels-server/efi64.bin ~# yum update