Problems configuring multi-chassis LAG with STP to core router

Question

I am trying to configure two M7100-24X in a configuration that looks exactly like this http://i.imgur.com/huVN2h9.png, ready to take to our datacenter and migrate our existing hosts onto it.

The datacenter provide us only with two links, a gateway address and tell us to use STP (they can't/won't aggregate these links at their end).

My test platform for their equipment is two links terminating at a bridged interface on a pfsense box with STP configured.

The issue is that flows arriving at the switch with the backup link (i.e. the one blocked by STP) seemingly won't cross the bridge to get to the gateway (or the other switch, if attempting to access the management interface, meaning one of the switches is inaccessible from the router side).

Flows between interfaces assigned to a VPC are rerouted and work as they should under all faliure modes, so it's seemingly just flows that need to cross the bridge that don't terminate at an MLAG interface that are affected.

STP is enabled on all ports and LAG interfaces, router has highest STP priority, one switch has higher priority than the other and the correct ports are designated as the next hops. Non-MLAG terminated flows just won't cross the bridge.

Can someone shed some light on:

a) If this is actually possible on this hardware?

b) If it is, what is the suggested configuration?

Thanks.

Retired_Member · Answer

Hi connorzc,
&nbsp;
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;Welcome to the community!&nbsp; &nbsp; &nbsp; &nbsp;&nbsp;
&nbsp;
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;I have tested the same scenario as you, the traffic(downstream/upstream) is ok.
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;So I suggest you check the configuration of the MLAG Switch, it is best to&nbsp;try and isolate the issue first.
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;A reminder:&nbsp;STP mode should be disabled on MLAG peer-link port, for the detailed information for MLAG feature and Admin guide, please refer the KB as below:
&nbsp; &nbsp; &nbsp; &nbsp; 1.&nbsp;MLAG on M7100--Basic setup and best practice
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;https://kb.netgear.com/app/answers/detail/a_id/25097
&nbsp; &nbsp; &nbsp; &nbsp; 2. Software Administration Manual: MLAG configuraiton, on Page73
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;http://www.downloads.netgear.com/files/GDC/M5300/M5300-M6100-M7100_SWA_v11_30Oct2015.pdf
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;
&nbsp;
Thanks,
EricNETGEAR® Community Team

connorzc · Answer

Hi Eric,&nbsp;Thanks for your reply.&nbsp;I appears that I am on a newer firmware than that manual was written for.&nbsp;I have consulted the manual for the latest firmware (the version I have installed), followed the configuration instructions as before and I am still facing the same issue.&nbsp;Let's simplify the situation, as I get the feeling it requires the same solution.&nbsp;I now have a setup which looks like the following:&nbsp;dev1 ----- sw1 ===== sw2 ----- dev2&nbsp;The sw1-sw2 peer link has been set as a LAG that was fully functioning before configuring the peer link.&nbsp;I have also disabled STP on the LAG and member ports, despite the current manual suggesting that STP across the peer link is actually possible.&nbsp;Now I am unable to ping across the link between dev1 and dev2&nbsp;which are connected to plain (i.e. non-MLAG configured) ports.&nbsp;Why isn't non-MLAG traffic able to cross the peer link?&nbsp;Presumably this is the same reason why in my previous post MLAG to gateway flows were not working depending on which switch they were sent to first.&nbsp;

Retired_Member · Answer

Hi connorzc,
&nbsp;
&nbsp; &nbsp; &nbsp; Thanks for your response immediately.
&nbsp;
&nbsp; &nbsp; &nbsp; &nbsp;As you saied, you give us one simplistic&nbsp;situation, but it's a LAG situation, not same as MLAG situation. And the configuration is not same too.
&nbsp; &nbsp; &nbsp; &nbsp;In LAG situation, all member ports need to enable STP to prevent the loop.
&nbsp; &nbsp; &nbsp; &nbsp;Then you said&nbsp;unable to ping across the link between dev1 and dev2, I suggest you can check LAG status at first. To confirm whether the LAG status is up or not.&nbsp;
&nbsp;
&nbsp;
Thanks,
EricNETGEAR® Community Team

connorzc · Answer

Hi Eric,&nbsp;The set up is the same, just with different types of links attached to it.&nbsp;From my first post, the main issue is where traffic that doesn't terminate at an MLAG won't cross the peer link.&nbsp;Here is a reference diagram to go along with the various use cases I am trying to demonstrate below:&nbsp;https://i.imgur.com/PRe9lv2.jpg&nbsp;dev-3 &gt; dev-4: works under all failure conditions where there is still a viable pathdev-4 &gt; dev-3: works under all failure conditions where there is still a viable pathdev-1 &gt; dev-2: doesn't workdev-2 &gt; dev-2: doesn't workdev-1/2 &gt; dev-3/4: works as long as dev-3/4.eth0 is updev-3/4 &gt; dev-1/2: only works if bond0 hashes the flow down the link to the switch directly attached to the recipient e.g. dev-3/4.eth0 &gt; dev-1 or dev-3/4.eth1 &gt; dev-2dev-3/4 &gt; dev-5 only works if bond0 hashes the flow down the link to the switch which has it's link to dev-5 as the designated STP root portdev-1/2 &gt; dev-5 only works if the common switch's has it's link to dev-5 as the designated STP root port&nbsp;I see no activity that suggests a loop has been formed at any point. The switches correctly designate STP root ports to create a non-looping path to dev-5.&nbsp;I hope this helps describe exactly the issue I am having.&nbsp;Connor

Retired_Member · Answer

Hi connorzc,
&nbsp;
&nbsp; &nbsp; &nbsp; Per your testbed topology, I have some question as below:
&nbsp; &nbsp; &nbsp; 1. Have you create LAG with port eth0 and eth1 on dev-5? If so, you should config VPC3 on two MLAG switches;
&nbsp; &nbsp; &nbsp; 2. If dev-5 does not support LAG, you should enable VRRP on two MLAG switches to Load Balance the traffic, and add all ports to same L3 VLAN interface, then set IP address for them;
&nbsp; &nbsp; &nbsp; 3. You said dev-1 to dev-2 not work, do these two dev in same subnet or different subnet? Then we can confirm the traffic should accross via L3 or L2 network.
&nbsp; &nbsp; &nbsp; 4. Would you please provide the configuration file for all switches? I think it will be helpful for us to debug.
&nbsp;
Thanks,
EricNETGEAR® Community Team
&nbsp;

Forum Discussion

Problems configuring multi-chassis LAG with STP to core router

5 Replies

Related Content

XS700 Series - MC LAG / MLAG (Multi Chassis LAG) support

Prosafe S3300-28X - Real multi-chassis (stacked) LACP possible ?

Time Configuration problems on a GS724Tv4

Nighthawk CAX30S LAN Configuration

Static IP Address Configuration for AP Mode

NETGEAR Academy

ProSupport for Business