Proxmox VE Changing Pass thru Disk Ids to qcow2

[tman]

Proxmox version: 6.4-13

Hello. I seem to have an issue with Proxmox turning or displaying my passed through hard disk to qcow2 format in my VMs. So for instance, I have 3 Proxmox VE setups. All three used to look like the below. Now only 1 Proxmox instance looks like below image.



Proxmox instance 2, I recently added another passed through disk (sata2). As you can see, from below image, the new disk is passed through to my VM. The VM using sata2 see it by its disk Id. Note however sata0 and sata1, both disk used to be listed by disk-id. Now they are listed as qcow2.


One thing I did notice is that I have Backups turned on in my Datacenter. In the VM all disk were selected for Backup. Would this be the cause for the qcow2 transformation? If so, how would I get the disks to be listed with disk-id and passed through format? Thanks in advanced.

 

[tman]

Can no one tell me what they think is going on here? This to me is quite alarming. I have no idea where to look for my data. Proxmox VM says these qcow2 files reside in my VMS storage, yet that is not possible due to the fact my VMS storage doesn't have the storage space to store 900GB.

 

[dcsapak]

pve will not rewrite the config to change the disks on its own (if it would, thats a bug)
can you check the task history if the config was changed (beside your addition of sata2) ?
do the qcow2 files exist?
did you restore the backups you mentioned?

 

[tman]

pve will not rewrite the config to change the disks on its own (if it would, thats a bug)
can you check the task history if the config was changed (beside your addition of sata2) ?
do the qcow2 files exist?
did you restore the backups you mentioned?

First of all, thank you for your response!

So what I know so far after trying to follow config trails, the raw/qcow2 files do exists in the storage that Proxmox assigned. Apparently however, these raw/qcow2 disk images are sparsed. So that is why they are not overrunning the local space. However in due time I can see local storage being overrun so I now must move the disks to a bigger storage area.

I'm still don't understand how the move happened. But my theory is that I restored the VM image from a backup. The physical disk were marked for Backup so Proxmox recreated the storage as raw/qcow2 upon restoring the VM. My issue now seems that these raw/qcow2 disks are being passed to my VM as RAID1 member drives within my VM which took the place of my physical disk. Now my question is how to move to a different storage without possible corrupting the RAID1?

So if this scenario is indeed the case, I might try to reallocate my physical disk to a ZFS RAID1 in Proxmox VE and move VM assigned disk to the ZFS storage.

 

[dcsapak]

But my theory is that I restored the VM image from a backup.

yes if you restored a backup, it will create new volumes for the backed up disk images. in the case of physical passed through disks we do not restore onto them.

My issue now seems that these raw/qcow2 disks are being passed to my VM as RAID1 member drives within my VM which took the place of my physical disk. Now my question is how to move to a different storage without possible corrupting the RAID1?

if the raid is inside your vm, just shut it off, move the disks, and turn thevm back on?

is there a special reason you want to directly pass through the disks to the vm?
most of the time it is much simpler to let the host manage the physical disks and only use virtual ones for the vms

 

[tman]

is there a special reason you want to directly pass through the disks to the vm?
most of the time it is much simpler to let the host manage the physical disks and only use virtual ones for the vms

1. Performance. I don't have top end hardware.
2. Migration or total failure recovery.

If the above is not the case with PVE, then I'll have to figure out how to manage raid disk using PVE and passing it through. How to?

• Do I create a new ZFS pool using RAID1?
• Do I pass in a single virtual disk to my VM since it is being allocated from a storage pool that is RAID'd?

However in my case to do the above, I wonder if could just move one of the VM current disks to the new ZFS storage, kill the raid within the VM and assign a single disk. All in the hopes of not losing data.

Your thoughts would be and are appreciated.