Wireless troubles

[BjornAx]

Hello!

I wanted to ask for guidence here as i have an issue at my work with our WiFi.

Some of our users (Not all of them) have issues in certain conference rooms when having meetings over teams.

The problems they have is that during the meetings the call can freeze and you see the "Trying to reconnect Call" message in teams and then 20 seconds later it's all fine.

What the users have seen when this happens is that the WiFi icon in windows drop from 3 to 1 bar and then it go back up to 3 again and everything is alright for the rest of the meeting.

This however is not a windows specific thing as employees with Mac and Linux have reported the issues as well.

Unfortunately when i have tested this in the specific conference rooms everything has worked as it should.

The only thing i have noticed is that when the issues occur the machine is connected to a AP that is where the user was sitting before the meeting and the machine then roams to a nearby AP. Why this would cause a 20 second freeze in teams i do not know however.

I have tried to change the transmit power to see if the computers will roam earlier but doesn't seem to make a difference.

What i have tried:

Changed AP's from 6 year old devices to new CW9162I AP's

Slightly changing the location of the AP's to try and not make them overlap too much

Doing a netspot survey and it said everything was fine from what it could see

Changing the roaming aggressivenes setting on the Windows machines.

Disabling 2.4 GHz and only allowing 5&6 GHz

Changing radio transmit power to 14-17dBm for 5 GHz and 17-20dBm for 6GHz

Updated WiFi drivers on the Windows machines

Added traffic shaping for the teams services with the Expedited forwarding DSCP tag

Sorry for the long text but i am a bit out of ideas. We don't have a large office and use 6 AP's. We have our office in a new wooden based building so no concrete walls that block signals either.

We use MX95 as the main router and then MS225-24 for the switches

[rhbirkelund]

Go to Network-wide > Clients, and click on one of the clients that have been having issues. From there select the "Roaming" tab, and scroll to around the time that they had the issue. Do you see the client roam? Perhaps, when the users leave their workspace to go to a conference room and have a meeting, they may be "sticking" to their old AP before roaming to a new AP that may be better, from the conference room point of view. 

Bad roaming between APs may contribute to the clients roaming from AP to AP. If roaming isn't seamless there may be some issues with coverage or transmit power. In the ned, it's the client that determines when it is time to roam. 

[BjornAx]

Hello!

Thank you for the suggestion. I have looked at the roaming logs and during the time the user reported the issue i can see that the computer roams between 3 different AP's. But the roam time is at the maximum 100ms so i don't understand how that could cause a 20 second freeze in teams.

Or is teams that unoptimized so it can't handle a 100ms roam?

Yeah i know that it's up to the client when it wants to roam unfortunately. 😞

I might have to do some more troubleshooting with the transmit power. I think our coverage is optimal at least netspot thought so when i did the survey.

[rhbirkelund]

I also tend to QoS mark all packets on the Switch settings to "Trust Incoming", as I seem to recall that teams does some QoS marking aswell. So besides the QoS marking on the Access Points, but also on the switch in order to preserve the markings through out. You may also want to do so on the MX. But remember that once the packet leaves the MX for the Internet, all QoS markings are at the mercy of the Internet. 

[cmr]

Have you tried allowing the dashboard automation to set AP radio power levels to see what that does? 

[BjornAx]

Yeah i tried that as well but still the same issue unfortunately.

[BlakeRichardson]

Firstly thank you for giving all those things a try, it's refreshing to see someone try and troubleshoot before running for help.

1. How congested is the area with neighbouring SSIDs? 

2. What method of authentication are you using, is it possible that EAP is timing out and the user is having to re-authenticate to the wifi? 

3. Do you have any traffic shaping enabled i.e. per user or per SSID bandwidth limit and could you be maxing out your WAN connection? 

4. Are all of the users joining the same teams meeting from Multiple devices? Could you try another approach of using a room based system reducing the load to a single connection? 

[BjornAx]

Yeah i try to solve my problems myself first so i can learn and do better before i ask for help.

1. Not very much. We are at the top floor in our office building and the only one on that floor. So no other SSID's on our floor but there are a couple of other on the floor below us. When we look at the SSID's on a windows machine i think we can see 2 or 3 others except from our own.

2. We are using a local radius server to authenticate against. I don't see any EAP timeouts in the logs and the users say the signal comes back without them having to do anything.

3. The only traffic shaping we use are the ones i activated to try and solve this so adding the teams services to expedited forwarding. No other traffic shaping is in place. 

I thought about the maxing out the WAN link as well but in the meraki dashboard it does not look like we are doing that. We only have a 1 Gbps link but it looks like we are not using all of it at any time.

4. Yeah multiple people can connect to the meeting even when in the same conference room. We have a external TV and sound in the conference rooms so one employee connects to that and share his screen to the others. We have it in our request list to buy room based systems but it has not been approved in the budget yet.

[PhilipDAth]

Since it is happening to everyone at the same time, we need to consider common factors.

First, your Internet connection.  Plug one user in with a wired connection.  Do they also experience the issue?  If so, examine your firewall and Internet connection.  It's not a WiFi issue.

When you are seeing this issue - are all the users connected to a single AP?

Have you tried turning off client balancing in the wireless profile?  Perhaps the AP is trying to force people to another AP because it is overloaded.

https://documentation.meraki.com/MR/Other_Topics/Client_Balancing

How many users are in these meetings at a time?

Have you disabled 2.4Ghz, and set the channel width for 5Ghz to 20Mhz?  Better yet, create a brand new RF profile using the wizard.  It uses best practice settings.  Just turn off 2.4Ghz for your main SSID.

https://documentation.meraki.com/MR/Radio_Settings/RF_Profiles

"Open Office" suits most office environments.

[BjornAx]

It's not happening to everyone unfortunately. Or at least not everyone has complained about it. 🙂

When they use wired connection everything seems fine. No freezes then from the testing we have done.

Client balancing is off in the settings.

Everyone that is in the same conference room is connected to the same AP but it's never more than 10 people in a meeting och maybe a maximum of 20 people connected to a single AP at any time.

2,4GHz is disabled on our SSID's. I have not tried to change the channel width so that could be worth a shot and try.

I could try and do another SSID for testing but most of the users that have this issue is not very technical, so they just want everything to work. 😄

I could always push out an intune configuration to one of the users with a SSID override.

[cmr]

@BjornAx 20 people connected to a single radio of a single AP all using Teams would be a potential issue.  I'm guessing that only some of the 20 are actually in the call.  The general metric for good wireless performance on 2.4 and 5 GHz is 30 devices per radio for data services and 10 per radio for real time services.