Random kernel panics possibly related to SCSI/VirtIO Block

[eexodus]

I have a few Linux VMs that get random kernel panics--approximately one panic every week in a group of 20 VMs. Searching for Proxmox raidx_tree_lookup kernel panics results in this thread that discusses VirtIO SCSI. These 20 VMs are configured as SCSI Controller = VirtIO SCSI and hard disk Bus/Device = VirtIO Block. Looking over the wiki am I incorrect in configuring them as VirtIO Block and instead should configure the Bus/Device as SCSI and not VirtIO Block? If so, is there any way to convert the disks to SCSI from VirtIO Block?

I have a few Windows VMs that never BSOD or panic--they are configured as SCSI Controller = VirtIO SCSI and hard disk Bus/Device = VirtIO Block. Should I SCSI on those as well over VirtIO Block?

 

[wolfgang]

Hi,

If so, is there any way to convert the disks to SCSI from VirtIO Block?

Yes, this is possible.

I personally would do a backup and snapshot before I start.

For Linux VM just detach the vdisk when the VM is offline and attach (double click on it) it back with scsi bus.

With Windows, I would recommend an assistant vdisk.
1.) create a new small disk with a scsi bus and attach it to the Windows VM.
2.) install the virtio scsi disk driver.
3.) shutdown the Windows VM.
4.) detach the virtIO vdisk and attach (double click on it) it back with the scsi bus.
5.) detach and destroy the assistant vdisk.
6.) start the VM.