[SOLVED] Bond interfaces

cbarbie

Member
Aug 24, 2021
7
1
23
I have configured several hundred bonded interfaces in the past, but now I am stumped. I have reviewed the guidelines again, but I must be missing something. Below is my lab config. After several failed attempts, I added an additional ip on nic2 to allow me access, as the bond config was not working. If I disconnect nic2, I can not access the server. Any help would be appreciated.

From the console, I can see messages when I connect/disconnect the different network lines.

This is on a fresh install - 9.2-1


Code:
auto lo
iface lo inet loopback

auto nic0
iface nic0 inet manual
#Col 1 - bot - 1g

iface nic1 inet manual
#Col 1 - top - 1g

auto nic2
iface nic2 inet static
        address 192.168.1.104/24
#Col 2 - bot - 1g

iface nic3 inet manual
#Col 2 - top - 1g

iface nic4 inet manual
#Col 3 - bot - 10g

iface nic5 inet manual
#Col 3 - top - 10g

auto nic6
iface nic6 inet manual
#Col 4 - bot - 10g SFP+

iface nic7 inet manual
#Col 4 - top - 10g SFP+

auto bond0
iface bond0 inet manual
        bond-slaves nic0 nic6
        bond-miimon 100
        bond-mode active-backup
        bond-primary nic6

auto vmbr0
iface vmbr0 inet static
        address 192.168.1.197/24
        gateway 192.168.1.1
        bridge-ports bond0
        bridge-stp off
        bridge-fd 0
 
Last edited:
Hi @cbarbie ,

There is not enough information in your post - you did not explain what the failure is, not even the Layer its at.

The best approach would be install tcpdump, if you don't have it yet, and trace each leg. Start with single physical link connected, check basic connectivity, confirm with tcpdump, plug the second port, remove the first one - continue checking. The idea is to check one at a time, then build the bond up.
Since you are using A/B type, I suspect you have a single switch in play - could this unknown switch be upset about MAC "flapping" from port to port?

Keep in mind - under the hood the PVE is a standard Linux OS (Debian/Ubuntu based). There are no special PVE knobs to turn here.



Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
Thanks for the tcpdump suggestion.

I figured out a solution:

1st attempts – no joy
• Install ProxMox using nic0 – 192.168.1.197
• Performed updates – no issue
• Network change
  1. Change vmbr0 to nic4 – did not apply
  2. Created bond0 with nic0 & nic6 – did not apply
  3. change vmbr0 to bond0, no change to IP – clicked apply
  4. lost connection to server
  5. tcpdump showed nics were active and seeing traffic


What worked:
• Install ProxMox using nic4 – 192.168.1.107
• Performed updates – no issue
• Network change
  1. Created bond0 with nic0 & nic6 – clicked apply
  2. Change vmbr0 to bond0 and 192.168.1.197– clicked apply
  3. lost connection to server on .107 but connected on .197
  4. removed config for nic4
  5. active/backup config worked as expected
 
Last edited: