VM image deleted itself? TASK ERROR: volume does not exist

[Neke]

Hi!

I run a small test infrastructure on PVE 7.0-8 with Windows Server 2022 and Windows 10 21H1 based OS. Today I decided to shirnk resources given to my Windows server OS, I changed core count from 4 to 2 and RAM available from 12GB to 6GB and proceeded to shut down OS(from OS interface). After veryfying that changes took place I proceeded to start my VM again, but got next error:
TASK ERROR: volume 'hdd-img:101/vm-101-disk-0.qcow2' does not exist

And my disk reports 60 GB less used space after machine shutdown:


I'm sure that I didn't even touched storage configuration and I have no idea how this happened which is VERY troubling. I did remove other VM's prior to this accident, but not this one and I was doing it with GUI, so no typing mistakes or anything alike. I posted screenshot of VM in question hardware in attachment. Any ideas what might have caused such behaviour?

I'm kinda new to PVE and it is possible that it was my fault at some point, but I really want to know why it happened...

 

[fabian]

what kind of storage is 'hdd-img'?

probably
- it's a shared storage used in multiple clusters without separation, and the VMID had a collision and the other cluster deleted it (don't use shared storage across clusters unless you have some sort of separation between the clusters - e.g., one subdir/export per cluster)
- or someone accidentally deleted the file at some point, but the deletion only became visible later on when the qemu process exited (while it is running it has the file open, so the data can't be deleted yet so it still shows up as used).

 

[Neke]

what kind of storage is 'hdd-img'?

probably
- it's a shared storage used in multiple clusters without separation, and the VMID had a collision and the other cluster deleted it (don't use shared storage across clusters unless you have some sort of separation between the clusters - e.g., one subdir/export per cluster)
- or someone accidentally deleted the file at some point, but the deletion only became visible later on when the qemu process exited (while it is running it has the file open, so the data can't be deleted yet so it still shows up as used).

Thanks for quick reply!

HDD-IMG is locally installed 1TB HDD, defined as Directory and used only to store images of VMs. There are no clusters, it's just a single PC station with PVE installed.

 

[Neke]

Okay this is stupid. It happened again with another VM on another physical disk. I guess I know now why Proxmox is never seen as productive virtualization platform.

I added new disk by following this guide:
https://nubcakes.net/index.php/2019/03/05/how-to-add-storage-to-proxmox/

And never after I touched server console, as well as never touched "Delete/Remove" button in GUI. Still another VM on another physical drive is gone after Stop/Start sequence. Good job.

 

[random nerd]

Okay this is stupid. It happened again with another VM on another physical disk. I guess I know now why Proxmox is never seen as productive virtualization platform.

I added new disk by following this guide:
https://nubcakes.net/index.php/2019/03/05/how-to-add-storage-to-proxmox/

And never after I touched server console, as well as never touched "Delete/Remove" button in GUI. Still another VM on another physical drive is gone after Stop/Start sequence. Good job.View attachment 28744

Same problem here. I was quite confused at first.... Leaving this here in case anyone else has the same issue.

I think perhaps there was a misunderstanding for both of us.

1) You correctly removed the INSTALLATION ISO.
2) You did not tell the VM to not look for the ISO when it boots.
2.1) This is exactly like a physical machine with the CD/DVd drive set to be the boot drive, exept a physical machine will usualy failover to the next boot device.

TO fix this I played around a little -

Datacetter -> PVE (you may have another name for it) -> VM number EG 101 -> HArdware -> CD/DVD drive (you may have multiple entries like IDE1 & IDE2) -> select "Do not use any media". Click OK and try boot your VM.

All of mine work again (Win/Ubuntu/etc)

 

[bcsw]

Okay this is stupid. It happened again with another VM on another physical disk. I guess I know now why Proxmox is never seen as productive virtualization platform.

I added new disk by following this guide:
https://nubcakes.net/index.php/2019/03/05/how-to-add-storage-to-proxmox/

And never after I touched server console, as well as never touched "Delete/Remove" button in GUI. Still another VM on another physical drive is gone after Stop/Start sequence. Good job.View attachment 28744

Agreed. Proxmox is NOT ready for production. I've had FOUR out of seven disk images 'disappear' after a reboot. One of the machines was not even running. There was no issue booting the VMs themselves.

They are gone.

No logs. No warnings. Just bye.

Running... Reboot. Since we have to blame the problem on something else, it must be these crusty R720s.

I'm seriously glad I tested before NOT forking out for a subscription.