VM suddenly shutdown after a few minutes

[Itay1778]

Hi, everyone
I pass a sata SSD that already has Windows 10 installed on it to a VM on Proxmox.
And after Windows recognized the hardware again the system start up and everything looks alright... but when I log in to my user so after a few seconds / minutes the VM just shuts down. Like I pressed Stop to the VM in Proxmox WebUI
I look at the Logs when it happened and I think it's related:

Jul 03 20:25:27 I1-PVE QEMU[21949]: KVM internal error. Suberror: 1
Jul 03 20:25:27 I1-PVE QEMU[21949]: emulation failure
Jul 03 20:25:27 I1-PVE QEMU[21949]: EAX=0000b364 EBX=00000000 ECX=00000000 EDX=000000b2
Jul 03 20:25:27 I1-PVE QEMU[21949]: ESI=00000000 EDI=00000000 EBP=d778f3b0 ESP=d778f278
Jul 03 20:25:27 I1-PVE QEMU[21949]: EIP=00008000 EFL=00010002 [-------] CPL=0 II=0 A20=1 SMM=1 HLT=0
Jul 03 20:25:27 I1-PVE QEMU[21949]: ES =0000 00000000 ffffffff 00809300
Jul 03 20:25:27 I1-PVE QEMU[21949]: CS =3000 00030000 ffffffff 00809300
Jul 03 20:25:27 I1-PVE QEMU[21949]: SS =0000 00000000 ffffffff 00809300
Jul 03 20:25:27 I1-PVE QEMU[21949]: DS =0000 00000000 ffffffff 00809300
Jul 03 20:25:27 I1-PVE QEMU[21949]: FS =0000 00000000 ffffffff 00809300
Jul 03 20:25:27 I1-PVE QEMU[21949]: GS =0000 00000000 ffffffff 00809300
Jul 03 20:25:27 I1-PVE QEMU[21949]: LDT=0000 00000000 ffffffff 00c00000
Jul 03 20:25:27 I1-PVE QEMU[21949]: TR =0040 86065000 00000067 00008b00
Jul 03 20:25:27 I1-PVE QEMU[21949]: GDT=     86066fb0 00000057
Jul 03 20:25:27 I1-PVE QEMU[21949]: IDT=     00000000 00000000
Jul 03 20:25:27 I1-PVE QEMU[21949]: CR0=00050032 CR2=1d212000 CR3=db04c002 CR4=00000000
Jul 03 20:25:27 I1-PVE QEMU[21949]: DR0=0000000000000000 DR1=0000000000000000 DR2=0000000000000000 DR3=0000000000000000
Jul 03 20:25:27 I1-PVE QEMU[21949]: DR6=00000000ffff0ff0 DR7=0000000000000400
Jul 03 20:25:27 I1-PVE QEMU[21949]: EFER=0000000000000000
Jul 03 20:25:27 I1-PVE QEMU[21949]: Code=kvm: /build/pve-qemu/pve-qemu-kvm-5.1.0/include/hw/core/cpu.h:651: cpu_asidx_from_attrs: Assertion `ret < cpu->num_ases && ret >= 0' failed.
Jul 03 20:25:27 I1-PVE kernel:  sda: sda1 sda2 sda3 sda4 sda5
Jul 03 20:25:28 I1-PVE kernel: fwbr100i0: port 2(tap100i0) entered disabled state
Jul 03 20:25:28 I1-PVE kernel: fwbr100i0: port 2(tap100i0) entered disabled state
Jul 03 20:25:28 I1-PVE qmeventd[710]: Starting cleanup for 100
Jul 03 20:25:28 I1-PVE kernel: fwbr100i0: port 1(fwln100i0) entered disabled state
Jul 03 20:25:28 I1-PVE kernel: vmbr0: port 3(fwpr100p0) entered disabled state
Jul 03 20:25:28 I1-PVE kernel: device fwln100i0 left promiscuous mode
Jul 03 20:25:28 I1-PVE kernel: fwbr100i0: port 1(fwln100i0) entered disabled state
Jul 03 20:25:28 I1-PVE kernel: device fwpr100p0 left promiscuous mode
Jul 03 20:25:28 I1-PVE kernel: vmbr0: port 3(fwpr100p0) entered disabled state

Here is the VM config:

agent: 1
bios: ovmf
boot:
cores: 4
cpu: host
hostpci0: 01:00,pcie=1,x-vga=1
machine: q35
memory: 8192
name: Windows-10
net0: virtio=66:2D:A7:45:6E:58,bridge=vmbr0,firewall=1
numa: 0
ostype: win10
sata2: /dev/disk/by-id/ata-HFS500G32TND-N1A2A_EJ73N465010103B9L,size=488386584K,ssd=1#This is the drive
sata3: /dev/disk/by-id/ata-WDC_WD20EZRZ-00Z5HB0_WD-WCC4M1FZ42ZF,size=1953514584K
scsihw: virtio-scsi-pci
smbios1: uuid=9178e446-8732-4416-87eb-32d9baaebd06
sockets: 1
unused0: local-lvm:vm-100-disk-1
unused1: local-lvm:vm-100-disk-0
unused2: local-lvm:vm-100-disk-2
unused3: local-lvm:vm-100-disk-3
usb0: host=045e:0745
vga: none
vmgenid: 4afbea3b-fdd8-4a0e-96e0-3e462e945fd9

Thanks
Itay

 

[Dunuin]

A good indicator is the IO delay of your host:


Is the IO delay too high, then your SSD is struggeling and the CPU needs to wait for the SSD to continue. Normally with a SSD the IO delay should be below 1% while idleing and heavy stuff like backups can increase to 10% or something like that.

 

[Itay1778]

Ein guter Indikator ist halt noch der IO Delay deines Hosts:
View attachment 27283

Ist der hoch, dann ist deine HDD überfordert und die CPU muss auf die HDD warten und die HDD bremst alles aus. Über 10% würde ich persönlich schon als hoch ansehen. Mit SSDs ist der meist gegen Null und mit HDDs die nicht komplett ausgelastet sind hängt der bei mir meist zwischen 2 und 5%.

I would love English next time

But such a thing would just be very slow no? Could it crash the whole VM without a clear explanation?
But I have a hard time believing that this is the problem anyway because this is a reliable SSD. And it doesn't cause problems if I use it with Windows directly on the hardware (without Proxmox)...

 

[Dunuin]

Sorry, I was switching between the German and English sub forums but looks like my head did not. I have edited the previous post so you can unterstand it.

And it doesn't cause problems if I use it with Windows directly on the hardware (without Proxmox)...

You get alot of write amplification using virtualization that you don't get when installing Win bare metal.
I for example got a write amplification of factor 7 from guest to host. Another write amplification of factor 3 from host to NAND flash. And another factor of 2 if I do sync writes. So for sync writes I get a total write amplification of 42 (7*3*2) so for every 1 MB I synconously write inside the VM 42MB are written to the NAND of the SSD.
That is one of the points why it is recommended to buy enterprise grade SSDs. Consumer SSDs aren't build to handle that much write amplification or 24/7 usage.

What is your average and maximum IO delay?

 

[Itay1778]

What is your average and maximum IO delay?

0.78 is the maximum.
like I said I do not think it's the problem.
it's either something in Proxmox that caused a crash or something in Windows. Tomorrow I will try to install the same installation of Windows (I have an Acronis backup of this installation of Windows so it will be the same) but as image-drive and not as Virtio passthrough.
It's clear to me that 24/7 operation needs SSDs that are built to work 24/7 which is why the SSD will not work 24/7
Either I set the drives to go to sleep or I turn off the host.
It's more of a workstation than a server