Vm backup rotation?

[WND AG]

Hello,
We have a lot of virtual machines running and if we back up all of them at once it bottle necks our backup target volume and they take forever and slow down the whole system.
Is there a way to start the backup job on the first vm and when that is finished it automatically starts a new job on the next vm and so on?

 

[guletz]

Hi,

You can create a custom script who can do this task(ansible ?)

 

[dietmar]

Is there a way to start the backup job on the first vm and when that is finished it automatically starts a new job on the next vm and so on?

This is the default behaviour of vzdump.

 

[adamb]

This is the default behaviour of vzdump.

On a side note, maybe he has multiple nodes which is why he is seeing multiple backups at once?

 

[WND AG]

yes, we have multiple nodes (18 in total) and like this its difficult to manage a daily backup. we can see a lot of problems (mainly disk io delays) during the backup to a nfs share. if the backup would start one vm after one regarldes on which node they run, this would much better. is there any chance to change this behaviour of the backup task?

 

[sienar]

I'm not using Proxmox in a clustered scenario yet, so I'm not 100% sure how backup jobs operate in that situation, but I'm curious how many backup jobs/tasks you're running? If you have all your VMs in one job, does it start a backup of a VM from each node simultaneously, so you end up with 18 simultaneous backups running?

 

[tom]

You can use bwlimit to reduce the load on your NFS server - not exactly what you request but it helps to reduce the load on your backup target.