BACKUP VM : some VM of other cluster freeze

[Alex24]

Hello,

I use Proxmox 3.3-2 with a cluster of 4 servers.
The storage is a SAN.
I tried to backup with NFS, SAMBA and LOCAL and i get this strange problem, i hope to get a solution.

Last scenario :

Backup some of Windows VMs on Server 1-3 in snapshot - LZO and local backup.
Backup is completed with success but after backup, some VM are frozen and these VMs aren't on the same server of the cluster.
By example, I have several VMs crashed on Server 1-1, 1-4...

One week ago, i tested NFS backup with stopped VM, and i get the same situation, and cluster was instable.

Can you help me to understand the problem and find a solution please ?

Thank you.

 

[manu]

Hi Alex
Are the backups and the running VMS on the same physical network, and using the same NAS ? You might saturating your physical link here.

Can you draw up a raw schema of your Network infrastructure ? ( hint: http://asciiflow.com/ )
BTW, Pve 3.2 is not supported anymore: upgrade to 3.4 at least

 

[Alex24]

Hi,

Runnings VMS use a 10 Gbits/s network for the SAN and between servers of cluster.
We have tried to backup to a NFS server (link 1 Gbits/s - VMs stopped to backup) and to backup (snapshot) to an external hard drive connected on one of Proxmox servers.

 

[manu]

Hi Alex
I would like to stress that again : 3.1 is old ! I don't know which kernel version you have, but for instance VirtIO drivers were deemed stable starting at the around the 3.3 kernel version, so if you're using Virtio in that for your VMs you might be in trouble.
We also know that a lot of people got their KVM problems solved when upgraded to the 3.10 kernel of PVE 3.4.
If you're not sure about how to proceed the upgrade , we have this forum here, and commercial support available for guaranteed response time.

 

[Alex24]

Hi,

The ProxMox version is 3.3-2 and kernel version : 2.6.32-33-pve Sep 26 2014

We can upgrade Proxmox, but we need to prepare it before.

But excluding the fact we run "old" version, i don't understand why when we start a backup of a VM, the cluster is instable and VM of others servers of the cluster are impacted (freeze of VM, need to reboot the VM), have you an idea of the source of problem ?
We need to back up VM before to upgrade.

Thank you for your help.

 

[manu]

Hi
Did you configure to get a mail report of your backups ? ( you can enter en email adress in the BAckup tab)
Inside you should get a report of what went wrong during the backup.

Also please paste the output of /etc/cron.d/vzdump from one of your node having problems

 

[Alex24]

Hi,

# cat /etc/cron.d/vzdump
# cluster wide vzdump cron schedule
# Atomatically generated file - do not edit


PATH="/usr/sbin:/usr/bin:/sbin:/bin"


The backup of the VM is ok, log successful but others VM of the differents servers of the cluster are impacted : freeze, need to restart the VM and it is this point i don't know why, and i can't backup because it is a production environment and i can't disturb others VM.

Thanks for your help.