vifmgr.bcastDomainPartition - multiple ifgroups within the same broadcast domain

[sron]

We have connected additional interfaces to our FAS8040 running ontap 9.3P5

These new Interfaces are connected to 10G Ethernet switches the old ones are just 1G and connected to different switches but physicially its the same network of a large environment.

I created a new LAPC ifgroup and added the new ports , created vlans on the new ifgrp added them to the broadcastdomain ( each vlan has its own) and and migrated all interfaces.

Since I have added ports from different ifgroups into one broadcast domain I get periodically the following error in my event logs:

ERROR         vifmgr.bcastDomainPartition: Broadcast domain <BCDOMAINNAME> is partitioned into 2 groups on node <NODENAME>. The different groups are: {a0a-<VLANXY>}, {a0b-VLANXY}. LIFs hosted on the ports in this broadcast domain may be at the risk of seeing connectivity issues.

Its not clear to me why this is an error or might be a problem, obviously it seems to be a problem if ports from different ifgroups are added to the same broadcast domain?

Can someone help me to understand the trouble this might cause?

And if its an error why is it allowed to create such a configuration?

Thanks

[naveens17]

command broadcast-domain show returns the correct information for the broadcast domain

net port broadcast-domain show

This error is due to communication between two ports on the same broadcast domain that are not able to talk through the Layer 2 switch.  This is most commonly seen due to a vlan not being passed through a switch interconnect or the specific ports in the ifgrp are a part of different vlans. 

so check  that all ports and all ISL's have the vlan's associated to the ports in the ifgrp

[sron]

I have checked all ports and all vlans,
I also tested the reachablity from a host on the same vlan to all lif`s hosted on different ifgroups
I also tested lif migration between both ifgroups and it works flawless without interruption

All ports are healthy. 

Is the assumed communication error between two ports based on some test I can do manually?

[JFMartin_ESI]

Are willing to run some commands and post "anonymized" output here for us to look at ?

[xandervanegmond]

Since both ifgroups have the same vlan configured, you run the risk of a packet coming in on a0a-XY, being sent out over a0b-XY since that interface is configured for vlan XY as well.

This can lead to some pretty funky networking issues on your switches and clients.

Although it may work normally now, I would recommend either setting up a second vlan, or setting up two subnets/broadcast domains for a0a-XY and a0b-XY.

[sron]

sure, tell me what you need or like to see

[sron · Accepted solution]

The reason was a different setting in the distribution function, the port health monitor l2_reachability check seems to fail if the distribution function is set to ip 

After setting the distribution function on any ifgrp to port the error message was gone