Problem with vzdump backup tool - backup too long

[François H]

Hi all,

Proxmox VE 1.9

I have a server installed with several VM and all VM host a debian stable.
One of them is a problem of time during backup: almost 2 hours to generate the tgz


vzdump --quiet --node 1 --snapshot --compress --storage NAS_Maurienne --maxfiles 1 --mailto toto@toto.fr 315

315: Jan 25 02:40:02 INFO: Starting Backup of VM 315 (openvz)
315: Jan 25 02:40:02 INFO: CTID 315 exist mounted running
315: Jan 25 02:40:02 INFO: status = CTID 315 exist mounted running
315: Jan 25 02:40:02 INFO: backup mode: snapshot
315: Jan 25 02:40:02 INFO: ionice priority: 7
315: Jan 25 02:40:02 INFO: creating lvm snapshot of /dev/mapper/pve-data ('/dev/pve/vzsnap-ausar-0')
315: Jan 25 02:40:02 INFO: Logical volume "vzsnap-ausar-0" created
315: Jan 25 02:40:02 INFO: creating archive '/mnt/pve/NAS_Maurienne/vzdump-openvz-315-2012_01_25-02_40_02.tgz'
315: Jan 25 04:32:55 INFO: Total bytes written: 36242022400 (34GiB, 5.6MiB/s)
315: Jan 25 04:32:55 INFO: archive file size: 6.49GB
315: Jan 25 04:32:55 INFO: delete old backup '/mnt/pve/NAS_Maurienne/vzdump-openvz-315-2012_01_24-02_40_02.tgz'
315: Jan 25 04:32:57 INFO: Logical volume "vzsnap-ausar-0" successfully removed
315: Jan 25 04:32:57 INFO: Finished Backup of VM 315 (01:52:55)

This one don't have the same problème


vzdump --quiet --node 1 --snapshot --compress --storage NAS_Maurienne --maxfiles 1 --mailto toto@toto.fr 312

312: Jan 25 02:00:02 INFO: Starting Backup of VM 312 (openvz)
312: Jan 25 02:00:02 INFO: CTID 312 exist mounted running
312: Jan 25 02:00:02 INFO: status = CTID 312 exist mounted running
312: Jan 25 02:00:02 INFO: backup mode: snapshot
312: Jan 25 02:00:02 INFO: ionice priority: 7
312: Jan 25 02:00:02 INFO: creating lvm snapshot of /dev/mapper/pve-data ('/dev/pve/vzsnap-ausar-0')
312: Jan 25 02:00:03 INFO: Logical volume "vzsnap-ausar-0" created
312: Jan 25 02:00:03 INFO: creating archive '/mnt/pve/NAS_Maurienne/vzdump-openvz-312-2012_01_25-02_00_02.tgz'
312: Jan 25 02:33:39 INFO: Total bytes written: 22663106560 (22GiB, 12MiB/s)
312: Jan 25 02:33:39 INFO: archive file size: 17.07GB
312: Jan 25 02:33:39 INFO: delete old backup '/mnt/pve/NAS_Maurienne/vzdump-openvz-312-2012_01_24-02_00_02.tgz'
312: Jan 25 02:33:41 INFO: Logical volume "vzsnap-ausar-0" successfully removed
312: Jan 25 02:33:41 INFO: Finished Backup of VM 312 (00:33:39)


Do you have an idea to solve the problem?

Thank you for your help

 

[tom]

I do not see a problem here. how big are your VM disks?

 

[daveo]

I wonder if the issue isn't that for what looks like the same source and dst, one vps seems to be much slower at backing up?

If I have understood correctly I could only imagine there must be something else going on?

 

[yccj]

I feel your pain. My question is how did you get VM 312 to go so fast?

I run a mixed KVM and OpenVZ container backup and the KVM VMs average 14MiB/s, but I'm lucky to get 7MiB/s for the OpenVZ containers. I have tried a number of tricks from the forums here and elsewhere to get it up to 7MiB/s. Personally, I think the difference comes from the fact that with KVM containers it's backing up a single raw file, but with OpenVZ it is copying a huge collection of small files. Does VM 312 happen to hold mostly larger files?

 

[e100]

Maybe this will help with the speed?:
http://forum.proxmox.com/threads/5967-Speed-up-vzdump-using-pigz

 

[yccj]

I was looking at that thread, but wasn't sure if it would help much with openVZ. Probably worth testing at any rate.

 

[jollyscots]

Wouldnt it be simpler to just reduce the compression level?

Surely there is a compression level where gzip is able to pull more than FIVE MEGS A SECOND??!!

lol. Didnt anyone at proxmox first try this on their own machine?