vzdump error: guest-fsfreeze-freeze failed

[gkovacs]

Wen backing up some KVM guests from ZFS to NFS, vzdump gives the following error:

Spoiler

Feb 15 01:35:05 INFO: Starting Backup of VM 101 (qemu)
Feb 15 01:35:05 INFO: status = running
Feb 15 01:35:05 INFO: update VM 101: -lock backup
Feb 15 01:35:05 INFO: VM Name: win2k3
Feb 15 01:35:05 INFO: include disk 'virtio0' 'local-zfs:vm-101-disk-1' 42G
Feb 15 01:35:05 INFO: backup mode: snapshot
Feb 15 01:35:05 INFO: ionice priority: 7
Feb 15 01:35:05 INFO: creating archive '/mnt/pve/NFS/dump/vzdump-qemu-101-2017_02_15-01_35_05.vma.lzo'
Feb 15 02:35:05 ERROR: VM 101 qmp command 'guest-fsfreeze-freeze' failed - unable to connect to VM 101 qga socket - timeout after 35864 retries
Feb 15 02:35:15 ERROR: VM 101 qmp command 'guest-fsfreeze-thaw' failed - unable to connect to VM 101 qga socket - timeout after 101 retries
...

As you can see it takes exactly one hour until vzdump tries this freeze and fails many times, after that the backup completes in normal time.

It only happens to a few VMs, most of them are not affected. Any idea what causes this?

 

[wolfgang]

Hi,

You have enabled in the VM options the Qemu Guest Agent and it do not answer.
I don't know why may bee you have not installed it or it is not running?

You have 2 options.

First
install/bring Qemu Guset Agent in the VM up again.

Second
deselect the Agent flag in the VM options.

 

[gkovacs]

Hi,

You have enabled in the VM options the Qemu Guest Agent and it do not answer.
I don't know why may bee you have not installed it or it is not running?

You have 2 options.

First
install/bring Qemu Guset Agent in the VM up again.

Second
deselect the Agent flag in the VM options.

Thanks for clearing that up. Unfortunately, I have quemu-guest-agent installed in many of the Debian 7 VMs, yet they still produce the error on backup, so I had to deselect the flag in the VM options.

 

[mvrhov]

This one is happening to us also and only on some machines. I've updated the quemu agent and there is no change.
The mos annying thig with that is. that this is going on for an hour.. Can proxmox catch this and cancel the backup of that particular VM and then retires at the end, or it tries it only for 5minutes or so. This is greatly annoying because of that the backup is still running when people come to work and all the VMS on that hos feel sluggish because the backup is still in progress.

 

[fabian]

This one is happening to us also and only on some machines. I've updated the quemu agent and there is no change.
The mos annying thig with that is. that this is going on for an hour.. Can proxmox catch this and cancel the backup of that particular VM and then retires at the end, or it tries it only for 5minutes or so. This is greatly annoying because of that the backup is still running when people come to work and all the VMS on that hos feel sluggish because the backup is still in progress.

the timeout got bumped because of the opposite problem - people complained that vzdump did not wait long enough for the fsfreeze (which can take quite a while on some OS apparently). if you have a VM where fsfreeze does not work reliably, you can just unset the "use agent" flag? AFAIK it is not really possible for us to detect whether it is failing or taking long.

 

[Andrew Holybee]

Didn't want to create a new thread for this I have the exact same problem I deselected the flag for the agent and restarted the VM but its scratched out in red still says yes what am I missing?

 

[wolfgang]

Hi,

you have to stop the VM and then start again.

 

[Andrew Holybee]

Awesome your always my hero so a restart wasn’t long enough for it to take affect. I will try that tomorrow night.

 

[Andrew Holybee]

I shut down the vm and restarted and it fixed the issue for me thank you very much