Inconsistent LACP system ID

[skendric]

PROBLEM

I am configuring (4) NICs on a two-node AF200 9.9.1P6 cluster with LACP, to attach to a pair of Cisco switches (configured with vPC)

The (2) Cisco switches are emitting identical LACP System IDs, which is good

* Both Cisco switches:  00-23-04-EE-BE-D1

But each of the ONTAP nodes are emitting different LACP System IDs, which is bad

* On Node 1, system ID = 02-A0-98-D7-EE-08
* On Node 2, system ID = 02-A0-98-D7-F3-01

CONFIGURE

I created the ifgrp as follows:
network port ifgrp create -node ontap-n1 -ifgrp a0b -distr-func ip -mode multimode_lacp
network port ifgrp create -node ontap-n2 -ifgrp a0b -distr-func ip -mode multimode_lacp

network port ifgrp add-port -node ontap-n1 -ifgrp a0b -port e0d
network port ifgrp add-port -node ontap-n2 -ifgrp a0b -port e0d
network port ifgrp add-port -node ontap-n1 -ifgrp a0b -port e0f
network port ifgrp add-port -node ontap-n2 -ifgrp a0b -port e0f

ontap-n1::> network port show -ifgrp-port a0b

Node: ontap-n1
Speed(Mbps) Health
Port IPspace Broadcast Domain Link MTU Admin/Oper Status
--------- ------------ ---------------- ---- ---- ----------- --------
e0d Default - up 1500 1000/10000 healthy
e0f Default - up 1500 1000/10000 healthy

Node: ontap-n2
Speed(Mbps) Health
Port IPspace Broadcast Domain Link MTU Admin/Oper Status
--------- ------------ ---------------- ---- ---- ----------- --------
e0d Default - up 1500 auto/10000 healthy
e0f Default - up 1500 auto/10000 healthy
4 entries were displayed.

ontap-n1::>

IFCONFIG -V OUTPUT

Node 1: ontap-n1
a0b: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
[...]
laggport: e0f flags=1c<ACTIVE,COLLECTING,DISTRIBUTING> state=3d<ACTIVITY,AGGREGATION,SYNC,COLLECTING,DISTRIBUTING>
[(8000,02-A0-98-D7-EE-08,0032,8000,0007),
(7F9B,00-23-04-EE-BE-D1,8005,8000,0121)]
input/output LACPDUs: 189 / 145043
laggport: e0d flags=1c<ACTIVE,COLLECTING,DISTRIBUTING> state=3d<ACTIVITY,AGGREGATION,SYNC,COLLECTING,DISTRIBUTING>
[(8000,02-A0-98-D7-EE-08,0032,8000,0009),
(7F9B,00-23-04-EE-BE-D1,8005,8000,4109)]
input/output LACPDUs: 218 / 145072

Node 2: ontap-n2
a0b: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
[...]
laggport: e0f flags=0<> state=d<ACTIVITY,AGGREGATION,SYNC>
[(8000,02-A0-98-D7-F3-01,0032,8000,0007),
(7F9B,00-23-04-EE-BE-D1,8005,8000,0125)]
input/output LACPDUs: 124 / 144983
laggport: e0d flags=0<> state=d<ACTIVITY,AGGREGATION,SYNC>
[(8000,02-A0-98-D7-F3-01,0032,8000,0009),
(7F9B,00-23-04-EE-BE-D1,8005,8000,410A)]


I am referring to the following KB article to interpret ifconfig output: https://kb.netapp.com/Advice_and_Troubleshooting/Data_Storage_Software/ONTAP_OS/How_to_Troubleshoot_LACP_Issues_with_ifconfig_-v_in_ONTAP_9.2?mt-view=...

QUESTION

What am I missing here?  Is there something about the 'ifgrp' creation that I have forgotten?

--sk

Stuart Kendrick

[AlexDawson]

This is expected behaviour. It is necessary to create a channel group to each controller, and then ONTAP will take care of moving IPs between nodes during failover or in case both links to a controller fail

hope this helps!

[skendric]

It seems to me though that I have done precisely that:  I have a created an LACP channel (a0b) and then assigned (4) links to it (two from each controller).  I want to present to the switches a single LACP System Identifier + Key, so that they perceive the ONTAP cluster as a single entity.

Are you suggesting that I want (2) channel-groups, each containing (1) port from each controller?  e.g. something like:

network port ifgrp create -node ontap-n1 -ifgrp a0b -distr-func ip -mode multimode_lacp
network port ifgrp create -node ontap-n2 -ifgrp a0b -distr-func ip -mode multimode_lacp
network port ifgrp create -node ontap-n1 -ifgrp a0c -distr-func ip -mode multimode_lacp
network port ifgrp create -node ontap-n2 -ifgrp a0c -distr-func ip -mode multimode_lacp

network port ifgrp add-port -node ontap-n1 -ifgrp a0b -port e0d
network port ifgrp add-port -node ontap-n2 -ifgrp a0c -port e0d
network port ifgrp add-port -node ontap-n1 -ifgrp a0b -port e0f
network port ifgrp add-port -node ontap-n2 -ifgrp a0c -port e0f

--sk

[AlexDawson · Accepted solution]

No, your first setup was correct for the NetApp controller side.

From the cisco side, you need to create two channel groups, one per NetApp controller, each with two ports - one port from each of the cisco switches.

One final piece of advice is that if you plan to carry the VLAN used for the OOB management port SP + e0M over the port channel, it should be the native VLAN 

Hope this helps!

[skendric]

I see ... I had been looking at this sideways ... thank you!  Works now

--sk