9000 MTU size on 1GB interface

[brucexx]

W have 3 networks with 3 separate dual port for each on bonded interfaces:
1. PVE Cluster 1Gbps,
2. Client facing network for VMs 1Gbps
3. Ceph Public - 10 Gbps

I do have 9000 MTU on the 10Gbps connection for Ceph but my question is really if it makes sense to put the 9000 MTU on the 1 Gbps Cluster network ?
As far as I can tell the backups are running through this interface and I guess it would help if we had a higher MTU size.

Thank you

 

[wolfgang]

Hi,

If I get you right do you make your backups on the same network as the corosync network resists?
This can be a dangerous combination with HA. In the worst case, the cluster dies when you make a backup.

All storage networks will benefit on high MTU, but corosync do not.

 

[brucexx]

Let me just clarify I am using Proxmox backup to backup VMs to a external NFS storage.

I guess I am using the vmbr0 for backup as "default" network which is also my cluster/corosync network , I was not aware that backup (theProxmox backup) network assignment can be changed. I dont see any type of network option in /etc/vzdump.conf .

Is there a way, I would love to move it to a 10Gb network interface which is currently underutilized.

Thank you