PVE 8 Backups Hang Other Vms

G

GGodden

New Member
Oct 7, 2024
11
0
1
Hi People,

I have a proxmox cluster that I have just updated from PVE7 to PVE8 and have a proxmox backup server that I have also just updated to the latest version. Since doing this I am finding that when backups start all other vms on the node hang for a little bit. I am seeing error messages on some of my vms that they are struggling to access their disks.

Has anyone else seen this?

Thank you so much in advance

George
 
Last edited:
I have the same problem a few days ago I posted on the forum but I still haven't gotten any response
 
Kingneutron said:
Have you tried fleecing?
Click to expand...
I haven't actually no.

Somthing I have noticed is that the version of multipath-tools is not the latest but when I run apt-get install multipath-tools it says it is running the latest version and then spits out the version number 0.9.4-5 which I know to be from 2023.

For a bit of context to this I am running my VMs off a SAN which prior to the update to PVE 8 was absolutly fine but now backups of VMs running on the SAN are causing everything to hang, including proxmox its self. If I move the storage to the local storage of the proxmox node and then back it up I have no issue.
 
GGodden said:
version number 0.9.4-5
Click to expand...
That is the latest stable version available in Debian PVE8.

There was a fix made last year to the multipath-tools in PVE8. So, could you show output for the following:
Code: 
apt list | grep multipath-tools
 
gfngfn256 said:
That is the latest stable version available in Debian PVE8.

There was a fix made last year to the multipath-tools in PVE8. So, could you show output for the following:
Code: 
apt list | grep multipath-tools
Click to expand...
Hi there,

Here is the output:
Code: 
apt list | grep multipath-tools

WARNING: apt does not have a stable CLI interface. Use with caution in scripts.

multipath-tools-boot/stable 0.9.4-5~bpo11+1 all
multipath-tools-dbgsym/stable 0.9.4-5~bpo11+1 amd64
multipath-tools/stable,now 0.9.4-5~bpo11+1 amd64 [installed]

good to know that I am running the latest stable version
 
gfngfn256 said:
And it includes the fix for a bug regression, see here.
Click to expand...
hmmm okay so looks like it isn't multipath that is causing my issue. Do you happen to have any other ideas I can potentially persue that could be causing this?

Thank you so much for the help so far
 
Maybe try setting a tmpdir in /etc/vzdump.conf to local storage as shown here in the wiki. Look for the example at the end of those list of options.
I'm not completely sure what effect this will have on your scenario/use case, but it may be worth trying.

I understand that it maybe difficult for you to schedule backups "out of hours" of the VMs being used, since you probably require constant backups even during usage hours. But maybe you could get creative & have a PBS server as a VM on your local host to which you would target all backups, & then once a day out-of-hours that PBS VM itself will get backed up to your existing PBS server, although this will probably depend on your available space locally.

You may also want to check the PBS forum, as possibly what you are encountering is due to the PBS update. I don't use PBS.
 
gfngfn256 said:
Maybe try setting a tmpdir in /etc/vzdump.conf to local storage as shown here in the wiki. Look for the example at the end of those list of options.
I'm not completely sure what effect this will have on your scenario/use case, but it may be worth trying.

I understand that it maybe difficult for you to schedule backups "out of hours" of the VMs being used, since you probably require constant backups even during usage hours. But maybe you could get creative & have a PBS server as a VM on your local host to which you would target all backups, & then once a day out-of-hours that PBS VM itself will get backed up to your existing PBS server, although this will probably depend on your available space locally.

You may also want to check the PBS forum, as possibly what you are encountering is due to the PBS update. I don't use PBS.
Click to expand...
I'll give it a go but I suspect it won't resolve the issue, what I have found is the issue only exists when backing up a vm that has its hard disk on my SAN and it doesn't seem to matter if I back up to local storage or to my PBS. The SAN is connected via dual 25 gig direct conect fiber connections with multipathing setup.
 
You must log in or register to reply here.
Top Bottom