NFS server snapshots - am I missing something?

[chrisalavoine]

Hello gents,

I have a problem with my NFS storage server. It's an Ubuntu 10.04 server running NFS with pretty standard export options.

I have another NFS share which runs from the Proxmox host itself.

I see very different performance between the two. I would expect some degradation but not to the extent that I'm witnessing.

This is the log from the latest snapshot:

230: Aug 21 00:07:01 INFO: Starting Backup of VM 230 (qemu)
230: Aug 21 00:07:01 INFO: status = running
230: Aug 21 00:07:03 INFO: backup mode: snapshot
230: Aug 21 00:07:03 INFO: ionice priority: 7
230: Aug 21 00:07:04 INFO:   Logical volume "vzsnap-ess-prox-004-0" created
230: Aug 21 00:07:04 INFO: creating archive '/mnt/pve/snap-nfs/dump/vzdump-qemu-230-2012_08_21-00_07_01.tar.lzo'
230: Aug 21 00:07:04 INFO: adding '/mnt/pve/snap-nfs/dump/vzdump-qemu-230-2012_08_21-00_07_01.tmp/qemu-server.conf' to archive ('qemu-server.conf')
230: Aug 21 00:07:04 INFO: adding '/dev/VMDATA2/vzsnap-ess-prox-004-0' to archive ('vm-disk-virtio0.raw')
230: Aug 21 00:15:43 INFO: Total bytes written: 21474839040 (39.46 MiB/s)
230: Aug 21 00:58:44 INFO: archive file size: 4.47GB
230: Aug 21 00:58:44 INFO: delete old backup '/mnt/pve/snap-nfs/dump/vzdump-qemu-230-2012_08_19-00_07_02.tar.lzo' 
230: Aug 21 00:58:47 INFO: Finished Backup of VM 230 (00:51:46)

As you can see there is a big wait between the "Total bytes written:" line and the "archive file size:" line. Has anyone experienced this before?

If I do the same snapshot on my NFS share running on the proxmox host there is no wait between these two lines and the snapshot finishes in around 8 mins.

Thanks,
Chris.