Wifi Stops Responding When Many Devices Are In Motion

[luhiadmin]

I work in the IT department at a medium size high school (about 1000 students). During passing periods the wifi mostly stops working. My theory is that this is due to the huge number of devices that are all on the move simultaneously (roughly 2000 devices) and connecting/disconnecting from many APs over the course of the passing period.

Are there settings that could be adjusted to help with this issue? Does anyone else have this issue? If so, have you found a solution? Thanks.

[BlakeRichardson]

I have not seen this issue and have worked with schools for a long time. It's very likely that private MAC is the problem and your DHCP range is being exhausted.

Basically a student is connected to an AP and walks towards another building an in the process loses connectivity because of a dead spot when they get within range of another access point their device reconnects using a different MAC even though the original DHCP lease is still active in the background.

I saw this issue very quickly when private MAC was released on iOS devices, the resolution was to make sure the DHCP range was a lot bigger than needed ( I would recommend a class A subnet) and set the lease time to an hour. 

[luhiadmin]

Thanks for the response. I'm not running into any issues with subnets running out of IP addresses to assign. All APs broadcast the same two SSIDs (staff wifi, and general wifi), so roaming between APs does not trigger a new DHCP lease.  

I looked around in wifi settings more and have made some changes to Meraki's Wireless Access control -> 802.11r settings for the "reduced overhead when a client roams from one AP to another".

[Ryan_Miles]

Pretty sure iOS MAC randomization only happens per SSID the client joins or when a client does a "forget network" and rejoins. iOS 14 beta had the every 24 hour thing, but I can't find any evidence that made it into production after the beta. Nor do my Apple clients ever change their MAC (unless I do a forget network action).

And it wouldn't occur on roams on the same SSID anyway.

[PhilipDAth]

What authentication method are you using?  WPA2-Enterprise mode?  If so, have you enable fast roaming on your RADIUS server?

[alemabrahao]

• How many access points do you have?

[Ryan_Miles]

I found the Support case you opened and looked at the linked Org.

You have 3 SSIDs. Generic names used here.

staff using a /23 subnet

student using a /20 subnet

guest using a /23 subnet

Side note the config for the staff SSID VLAN tag doesn't make sense or it's an expired config. It points to a tag to use VLAN 10, but no AP has that tag. So it follows the default tag of 50. Not an issue, but a confusing config if an admin looks at it and thinks it's putting clients on VLAN 10.

While you're not running out of IPs you're possibly hitting limits of the MX250 upstream. A MX250 is recommended for use with jp to 2,000 clients per the MX Sizing Guide. I've seen plenty of instances in which DHCP and bandwidth is not exhausted on a MX, but far more clients than it's rated for are connected through the MX and flow capacity is being exhausted (traffic dropped). To verify that I would call into Support during your busiest time of the day/highest number of clients connected and have them check the flow count of the MX.

When your clients don't work what are the exact symptoms? Do they still show as connected to Wi-Fi, have an IP if you check settings, but just can't get to the internet? Or, are they actually disconnecting, looking for another SSID or AP to join?

[Ryan_Miles]

You're also having some WAN uplink failures. Today, WAN 1 has had brief failures at 8:31 and 9:04 local time. There have been failures in past days as well. And your WAN 2 has a much lower bandwidth setting (especially upload).

Just trying to add more context as this might not be a wireless issue. But it depends on what exact failures your clients are experiencing. 

[TBHPTL]

Ill bet L2 is just peachy....

[UKDanJones]

Knowing how the clients are disconnecting would be super helpful (no IP, no association, auth timeout) etc. 

a packet capture would be even better 😉

tou might be able to try downloading metageeks new channelizer 6 and running that to see what it picks up. 

[TBHPTL]

I didn't see anything stating how the SSID's were setup, bridged  mode or NAT mode. Also nothing with regard to which firmware you are running nor any mention of AP models etc...