Large VM Disk Imports Crashing Host

[Samuel-SSH]

Hi,

Whenever I use qm importdisk for any disk with a sizable amount of data on it, it absolutely blasts IO, making everything else on the server slow to a halt and sometimes crashing the server entirely. We're running RAIDZ1 with 3 2TB HDDs.

Does anyone know why this happens and a way to fix it or limit the IO qm importdisk uses?

Thanks.

 

[zartarr88]

same issue with me ; surprised you haven't received a response on this...
I have a 380gb qcow2 VM ;
i get a crash around the 130gb (33%) mark ; just hard resets.... entire proxmox server...

 

[Neobin]

How does your complete setup of all involved storages look like?
Hopefully not similar bad as this from the OP...

 

[zartarr88]

fast10 and zfsdata01 is a 2TB nvme ssd
seagate is a 8tb traditional drive (2 of them in zfs raid)
local / local-lvm is just where I installed proxmox on like a 256gb ssd

i used NFS to copy a qcow2 image to /zfsdata/images ...
then do qm import... crash crash crash

oh also... I had a disk attached to a VM and magically I'm missing it now as well... so somehow it just vanished without removing it !?

 

[Neobin]

What are the exact model numbers of those disks each?
How does your IO-delay look like during the import?
Something in the syslog before the crash?

 

[zartarr88]

I believe it had to do with the Aquantic AQC-107 NIC ; I just limited the speed to 1gig... not sure what else to do .