it steals space from the node during the backup

[frankz]

Hi everyone, I have noticed that in the last few days the LXC machines are no longer backing up. Looking closely at the type of error I noticed that in the system before backing up to PBS (in my case), a .tmp file is generated and the end is saved to PBS. This is probably due to the virtual disk in nfs. I noticed that it fails due to the increasing consumption of node space. Has anyone encountered the same problem?
I think the problem arises from the fact that the node tells me that the disk that resides on nfs does not support the snapshot so it saturates the disk of the node. Considering that I have several LXCs on nfs but smaller, it would be advisable to evaluate more root space at the node from the first installation?

 

[dcsapak]

hi, yes, if you do snapshot backups for container volumes that do not support snapshot, it does a 'suspend' backup
this first syncs everything to a tmpdir and backup at the end

you can configure the tempdir with the file '/etc/vzdump.conf'
there you can give a path where there is enough space

 

[frankz]

hi, yes, if you do snapshot backups for container volumes that do not support snapshot, it does a 'suspend' backup
this first syncs everything to a tmpdir and backup at the end

you can configure the tempdir with the file '/etc/vzdump.conf'
there you can give a path where there is enough space

 

[frankz]

hi, yes, if you do snapshot backups for container volumes that do not support snapshot, it does a 'suspend' backup
this first syncs everything to a tmpdir and backup at the end

you can configure the tempdir with the file '/etc/vzdump.conf'
there you can give a path where there is enough space

OK thanks . But if the LXC machine is in HA and therefore subject to movements between nodes, how can I, for example, declare a destination that may not be on the node? Maybe it should be a network share?

 

[dcsapak]

declare a destination that may not be on the node?

the /etc/vzdump.conf is local to every node, so you can have a different tempdir on each node

Maybe it should be a network share?

you could do this, but then the content of the container will be transferred over the network multiple times
1. sync local -> share
2. share -> local (during actual backup) -> backup server

 

[frankz]

the /etc/vzdump.conf is local to every node, so you can have a different tempdir on each node


you could do this, but then the content of the container will be transferred over the network multiple times
1. sync local -> share
2. share -> local (during actual backup) -> backup server

I apologize, but due to the translation I misunderstood or should I always edit the vzdump or something else? Thank you and excuse me