Layer3 roaming not working between controllers

[Craig Le-Butt]

Hi

We have several controllers on 1 site one, we seem to have a problem when roaming, the device getting a new IP.

We've since added a couple of mobility anchors for a couple of companies which are working no problems.

What I have noticed is that our own controllers in the local mobility aren't in the mobility anchor for the WLAN.

Do we just need to add local or all the wlc on that site to the mobility anchor for that WLAN?

Cheers in advance.

[Richard Atkin]

Hi Craig,

I'm not entirely sure I understand your problem.

Inter-WLC roaming is enabled just by having the WLCs all in the same Mobility Group.  Using a Mobility Anchor will force all traffic to be tunnelled to the Anchor.  If you have an inconsistent config where some internal WLCs are anchoring an SSID to a remote WLC and others aren't, this could cause you no end of problems.

[Scott Fella]

You need to provide more info... it is hard to understand what your question is.  Maybe post your config and maybe a diagram and what exatly is the issue, with clients and the ip address.

Thanks,

Scott

Help out other by using the rating system and marking answered questions as "Answered"

[Craig Le-Butt]

Thanks for the replies.

This is what has happened, one building has 200 APs in controlled by 2 WISM.

Floor 0-4 are one one wlc, floor 5 -6 on another wlc.

On floor 5 a COW is moving, but it is connecting to floor 4 and 5, but hen this happens the IP is changing on the COW.

It seems like the roaming isn't working, so upon checking the mobility anchor on the Ssid there is a mismatch between the 2.

One has the local and the address of a couple of the wlc on that it, the other wlc has nothing in it.

So as the wlc that is in the local mobility group, should there be an entry for the  the Ssid?

Not back on site till tomorrow.

Cheers

[Stephen Rodriguez]

You shouldn't have to anchor for that scenario to work.  you should just have to have all the WLC in the mobility list/group.

When the client roams to a new WLC, it sends a mobility message to the group to see if anyone is aware of the client alreayd, and either does a L2 roam if the interface subnet is the same.  Or a L3 roam if they are diffferent.

In the case of a L3 roam, the traffic is 'auto-anchored' to the original WLC, so you dont' have to worry about RPF checks.

HTH,
Steve

------------------------------------------------------------------------------------------------
Please remember to rate useful posts, and mark questions as answered

[David Watkins]

You don't need to have any "anchoring" configured given the description of your scenario, but they should be in the same mobility group.  From your OP it looks like you have half the floors on one WLC, the 2nd half on another.  It seems that the clients will hit APs a floor below (when operating on floor 5) causing a L3 roam, however your client is getting a new IP.  You mention also that the config is different.  What is most likely happening is that your client associates to a floor 4 AP, the WLC for that floor makes a mobility announcement, however the floor 5 WLC cannot handoff due to WLAN config mismatches.  This results in at least a 2 second outage, awaiting the mobility event from the other WLC, then the client has to associate as an entirely new client on the next WLC, thus getting a new IP and leading to a broken roam.

You will want to make sure your WLAN configurations are "identical" (other than interface in use) so that when a client on floor 5 roams to a floor 4 AP, the WLC can properly handoff the client to the next WLC and maintain it's existing IP and provide a more seamless roam.

[David Watkins]

Too fast for me Steve

What Steve said!