AP client roaming with only 1 AP

[szymonsikorski]

We have an issue on a site with an embedded WLC running on a 9300 series catalyst with only a single AP associated to it. For some reason the clients keeps roaming every 1-3 seconds despite there being only one AP broadcasting the SSID. The roaming aggressiveness has been turned down to the minimum. The same WLC configuration is running on all of our other sites where there are no problems. All users also have the same laptop configuration and wifi profiles on all sites. When a user from a different site comes to the site with the issue they can't connect either. Laptops from the affected site work fine on other sites. What can cause a device to choose to roam when there is only a single AP in the office?

[Flavio Miranda]

Hi

 Roaming would be the client moving from  one AP to another, which would be impossible as per your description.

 Either the clinet is moving from 2.4ghz to 5.0ghz radios  or they are disconnecting and reconnecting

 Either way a client debug on the wlc would be important to see what happens.

[szymonsikorski]

Yeah I agree, my gut feeling tells me its trying to move from 2.4 to 5 as well. We're going to firce the wifi adapter to stick to a single band and see how it goes.

[Rich R]

Don't start changing things until you understand what the problem is.
Do a debug on the client MAC address like Flavio said.  Once you've captured debugs when you observe the problem then run the results through https://cway.cisco.com/wireless-debug-analyzer/ to get a nicely formatted summary.
It could just as easily be interference causing your issue in which case you need to find the source.

Regarding 2.4 vs 5 - good idea to have devices prefer 5 where settings allow that.
Also disable lower data rates (but be aware that might reduce cell size/coverage).
And band-select configured on the WLC will help nudge clients to 5 GHz.

Also what model of AP?  Is it the same AP at other offices?
What version of software on 9300?  SDA (DNAC) or standalone mode?

[JPavonM]

Have you checked that the laptops have the last BIOS update and wNIC driver from the vendor? If so, what about testing with latest driver from the manufacturer? (If it's an Intel one, then you're lucky as Intel publish new drivers on it's web site for all wNICs, that maybe newer than those from the vendor.

Regarding your WLAN Infrastructure, why are you broadcasting the corporate SSID in both 2.4- and 5-GHz band simultaneously? All modern laptops support 5 GHz so best way would be to NOT broadcast this SSID in 2.4 GHZ as this is well-known to be overutilised and it's full of non Wi-Fi interference that may reduce the quality of experience a lot.

Using band steering is not a good option as it may happen that the devices keep trying to connect to the other band so traffic disruption may happen, specially with real-time applications such as conferencing ones.

[Rich R]

Yes that is the standard warning for band-select with real-time (voice/video) apps but we've not seen it cause any problems in practice.

[szymonsikorski]

After disabling 2..4GHz we're observing the same issue. The laptop connects to the SSID for a split second and windows event viewer shows a roaming notification in sub second intervals. With only 1 AP on site and just 1 range (5Ghz) enabled now we have no clue what it's trying to roam to. The WLC debugs are showing on obvious signs to go on and from it's perspective the user simply drops and tries to get on again. The latest wireless drivers are installed on all machines. We've had people in the office from another office which doesn't show the same issue and they aren't able to join the WiFi either. We're going to send out an engineer with their own laptop, a wifi survey tool and a brand new AP. All of our sites are configured identically and none are showing this issue. This strikes me as a signal issue between the AP and the client although when they do manage to connect for 1 second the WLC shows the signal strength and noise ratio and healthy. Something triggers the laptops to roam to another network but we have no clue what it could be either. The WLC isn't reporting any interference or rogue APs with the same SSID that the user could roam to.

[Flavio Miranda]

You dont have Cisco Prime dont you?

Prime have better log view then wlc directly

 One thing that could be is Meraki Aps closer to your AP. I got this porblem once. They were usimg the contention feature and sending deauthentication to my APs. User was disconnecting all the time.

I did not see it on the wlc but I could see on Prime.

 The solution was the Meraki guys disable the feature.

[szymonsikorski]

No unfortunately we've decommissioned prime as all sites are now SD-Access via DNAC. DNAC doesn't show anything like that though.

[Flavio Miranda]

If the wlc is mamaged by DNAC at least for assurance, Yes, you should be able to see just like Prime

 DNAC was suppose to be better then Prime. Take a look in Intelligent capture.

[Rich R]

Like @Flavio Miranda was saying you might be victim of containment (not contention) from a neighbour.  I've mentioned it before - we had a hotel customer whose neighbour got a new Meraki network and enabled containment without understanding what they were doing.  After we traced the problem and went over to have a chat with the neighbour and explained what they were causing and that it was illegal the "network admin" was very apologetic and turned it off.

[JPavonM]

As maybe all wireless engineers globally, we have been and are been impacted by untraied administrators that do not understand how containment works and that it's illegal in the majority of the scenarios and regions.

It is a hard work to find out if you are impacted by this, but using tools such as WiFi Hawk you can identify it easily looking into the clients table.