Legacy ubuntu VM killed by backup?

[dragon2611]

Any ideas for this one, Essentially I have a VM where doing a backup snapshot kills it.

PVE 6.1.5,

When doing a snapshot backup of a legacy ubuntu 14.04 VM to a NFS share provided by a centos8 VM onto a VDO volume backed by an ISCSI mount on an RS814 (The Centos VM doing VDO is excluded from the backup run) It will cause a kernel panic on the Ubuntu 14.04 VM.

Other VM's based on later kernels/OS seem to throw errors to the console about cpu stalls and I see a few minutes of disruption but they seem to recover themselves once the backup is done, it's just the Ubuntu 14.04 VM (Which I hope to replace soon) that actually dies.

I'm not worried about the other VM's and the CPU stalls I just consider that as par for the course due to using low-end hardware.

This is on hardware i've been able to acquire cheaply for personal use so nothing enterprise here.

 

[Stefan_R]

Could you try backing up to a non-NFS storage? (i.e. a local disk for testing, or SMB/CIFS etc...)

People have been reporting some issues with NFS hanging their VMs recently, could be that the kernel in 14.04 doesn't like that too much.

 

[dragon2611]

Providing there's enough room on the proxmox root partition I might be able to back it up there temporarily.
It always used to backup on when the storage wasn't NFS so I'd say that's a good bet.

Also the NFS VM has a local raid vdisk passed through to it, I might also try an NFS export pointing at that (Seems to be a bit faster in terms of write speeds/responsiveness)

I did do a manually triggered backup of the VM without it dieing last week so I suspect it's to do with the storage performance and how long it's stunned for.

Edit:

Trying running the full backup again to the same storage but with bwlimit set to 30MB/s , Noticed when I kicked of a backup the CPU for the storage VM went through the roof (Expected it's doing compression/duplication) and the iowait for the server overall was in the high 40-60% range.