ONTAP Discussions

ifgrps spanning two switches


Hi there,


I was looking for some opinions on the best approach to a situation I have.


I have a HA pair with each node having 4 ethernet ports. On each node, 2 of the four ports go to switch A and 2 go to switch B. We cannot use LACP across the switches. For redundancy and bandwidth purposes I created an ifgrp on each node that trunked them together. I set it up as a multimode ifgrp based on IP policy. The odd thing is if down both connections to switch A  (leaving just both switch B connections) the svm continues to work however if down both connections to switch B (leaving the A connections up) the svm doesnt ping. All interfaces are up in usual mode and its not like the ports on switch A appear to be faulty.


Therefore given the above scenario how would yourselves setup the networking for best balance of redundancy and bandwidth. I know i can make it easy and just go single mode but that seems a waste of links, also as mentioned we cant use LACP as the switches don't support it and we can group across the separate switches. The ports are just plain connections and I was hoping I could do the grouping on the storage side.


We are using CDOT 9.7, any advice you can give would be greatly appreciated.







Unless the switches understand the cross switch lacp (like Ciscos vPC or Arista LAG) then you cannot do all four ports in the same ifgrp across both switches. 


You could put one node on switch a and one node on switch b 


Then be sure to distribute your load between both nodes (like flexgroups)


You could also do this

Each node gets a 2 port port-channel to each switch. Then put two data LIFs on each node, one on each ifgrp. This will help distribute also. 


I'm sure there will be other suggestions. You just can't do what you're trying to do right now