Error 46, 41 and 6008 at each windows 2016 server start-up

[Frigg]

Hello, at each start up of a windows 2016 server VM, I have error event just after VM's start-up:
46 source volmgr: The initialization of the crash dump file failed.
41 source Microsoft-windows-kernel-power: The system restarted without stopping correctly. This error can occur if the system stops responding, has crashed, or is no longer powered unexpectedly.
6008 source Eventlog: The previous system shutdown at 08:22:53 on 23/08/2019 was not planned.

What happend is that PVE shuts down in the evening if VM has shutdown.
At PVE's boot, the VM boots at start.
There is only these VM that is starts at boot. So there shouldn't be any overload problem. But though the VM has started and is in use there is these errors although the VM doesen't restart, and users can work without any interruption or restart.


As proxmox manages VM's power, I told myself that the problem may comes from it.

To be complete, I have a HID button over interrupt driver error, but when I want to update the driver, windos says the latest driver is already installed



What can I do to fix these errors ?

Delay start at boot from 1 or 2 minutes ? If yes, how can I do that ?

Startup delay: Defines the interval between this VM start and subsequent VMs starts . E.g. set it to 240 if you want to wait 240 seconds before starting other VMs.
But in my case, it is the first VM to start. In other words, how can we delay the start of the first VM start at boot?


Thank you for your Help

Best regards
Hervé

 

[Stefan_R]

It seems that what is happening here is that when PVE shuts down in the evening, it issues an ACPI shutdown to the Windows VM, but since there is a driver issue on Windows, it doesn't receive the shutdown command. Thus, PVE runs into a timeout, and forcibly kills the VM to complete its own shutdown, creating the log entries you see.

You can try installing the QEMU guest agent in Windows, bypassing the broken device driver for ACPI. You can verify by clicking on "Shutdown" in the PVE Web GUI (or running qm shutdown <vmid>) and checking if the VM actually shuts down.

 

[Frigg]

Hello Stefan,
Thank you for your answer.
It can't be what you are assuming.
I run this script to check there is no VM running, to shutdown the node.
The VM shuts down with instructions given in windows.
If the VM hasen't shuted down, the node doesn't shuts down.



Qemu agent is already intalled and Enabled

Please note that windows gives this error dated few seconds after its boot. It is like if power supply was not enough for the node finishing to start, and Vm's start. That is why I'm asking if ther is a way to delay start at boot's first VM.
If it is not that, is there a way to trace what happens when booting linux?

Best regards

 

[Stefan_R]

You can see the boot process of linux in the syslog, use 'dmesg' to print it to terminal.

I doubt you will find your issue there though, since Windows (in the VM) cannot know anything about boot errors of the host. Are you sure Windows is being shut down correctly?

Please note that windows gives this error dated few seconds after its boot.

The errors indicate that Windows has crashed, so for obvious reasons it was unable to write the log files back then (since it crashed). It messages you on the next bootup, to notify that the last time the system was turned off, it didn't happen cleanly. This doesn't mean that the error happens on boot.