1

I have 2 switches and a server with 4 NICs. The switches are connected together with a LACP LAG.

I'd like to have a situation where I'm able to survive a switch failure, also still have the ability to lose any of the links and get the benefits of increased aggregate bandwidth of traffic flows into each switch that a LACP bond brings.


  .-----------.   .-----------.
  |  Switch1  |===|  Switch2  |
  '----=---=--'   '--=---=----'
        \ /           \ /
        / \           / \
  .----=-.-=----.----=-.-=----.
  | eth0 | eth1 | eth2 | eth3 |
  '------'------'------'------'

Per this question https://unix.stackexchange.com/a/172232/30008 I could have a single mode 4 bond of all 4 NICs if I didn't have the inter-switch link.
I was struggling with that setup because bond0 takes the MAC address of the first link in the bond. And even though the bonding driver negotiates an active Aggregator ID (pair), the link status on the other pair is still up, which causes the passive switch to keep an entry for the bond's MAC.

The Linux bonding driver can't bond bonds, otherwise doing an active-backup bond across the LACP pairs would work nicely.

What's the best way to achieve this with non-stacking/mLAG switches?

  • And there is no chance you could break the LACP between switches? – user34720 Jan 15 '16 at 12:53
  • And replace it with a single non-LACP link you mean? Or get rid of the inter-switch link altogether? I'd prefer not to - I'd like to find out how best to set this up with that in place. – batfastad Jan 15 '16 at 14:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.