Use of NIC bond (bond0) on each PVE cluster network host node (retrospectively)

[tuathan]

I have a Proxmox cluster (5 host nodes) where each host nodes uses vmbr0 for network access (contents of /etc/network/interfaces includes):

auto vmbr0
iface vmbr0 inet static
        address 10.XXX.XXX.XXX/24
        gateway 10.XXX.XX.X
        bridge-ports eno1
        bridge-stp off
        bridge-fd 0

Cluster VMs on different host nodes then communicate via:

auto vmbr2
iface vmbr2 inet manual
        bridge_ports vxlan2
        bridge_stp off
        bridge_fd 0

auto vxlan2
iface vxlan2 inet manual
        vxlan-id 2
        vxlan_remoteip 10.XXX.XXX.XXX
        vxlan_remoteip 10.XXX.XXX.XXX
        vxlan_remoteip 10.XXX.XXX.XXX
        vxlan_remoteip 10.XXX.XXX.XXX

I'd like to retrospectively use a bond of eno1 and eno2 for host nodes.

1. Can I simply add the following to each host node?

auto bond0
iface bond0 inet manual
      bond-slaves eno1 eno2
      bond-miimon 100
      bond-mode 802.3ad
      bond-xmit-hash-policy layer2+3


auto vmbr0
iface vmbr0 inet static
        address  10.XXX.XXX.XXX
        netmask  255.255.255.0
        gateway  10.XXX.XXX.X
        bridge-ports bond0
        bridge-stp off
        bridge-fd 0

2. Is above bond0 / vmbr0 compatible with Proxmox cluster (i.e. if I do this on every host node)?
3. Is the bond-mode ok?
4. Can I add this retrospectively to each host node (even though already in cluster using vmbr0)?
5. Is the bond compatible with vmbr2 / use of vxlan?

Thanks in advance!

 

[aaron]

1. Can I simply add the following to each host node?

That should work

3. Is the bond-mode ok?

That depends on what your switches can do and how they are configured. A very basic mode would be the "active/backup" mode that should work always.

Make sure to have access to the nodes by other means than the network in case something goes wrong, so you can change back/fix the config.

 

[tuathan]

That should work

That depends on what your switches can do and how they are configured. A very basic mode would be the "active/backup" mode that should work always.

Make sure to have access to the nodes by other means than the network in case something goes wrong, so you can change back/fix the config.

I decided to do it while i was onsite today for that very reason! It worked no issue! Thanks.