WiFi client drops?

[jmhouse96]

If you look at the attachment from Prime you see the waves of clients connecting... We have 4800 series APs and 5520's running 8.8.100 code (I have also tried 8.8.111) and in Prime I see this as my daily connected clients. I do not get complaints saying people are getting disconnected, but it seems to me that something is happening here or Prime has an issue. I have a few TAC cases open but wanted to see if anyone else is seeing this.

[Jurgens L]

I've been running 8.8.114.132 for a 4800 deployment for a month on 5520's, so far don't see anything similar to what you are experiencing.

[Haydn Andrews]

As no one has complained could mean 2 things:

there is no issues and PI is reporting wrong

Users are not using real time applications and not noticing the drop.

As you are running 8.8 it would be possible to use the Webhooks and pull stats from the WLC regularly to see if the client counts are matching PI, or even do it manually.

I have had PI have reporting issues in client counts in the past, was a DB issue and TAC sorted it.

[Leo Laohoo]

If you are running 8.8.X.X and the clients are Intel chipsets, then this could be CSCvn66715.
If this is the case, then this bug should be fixed in 8.8MR2.
PS: If you are running in HA SSO, pay close attention to your memory buffer "sh memory mbuf summary". Make sure "Number of Mbufs In Use" is < 1000.

[Scott Fella]

What do you see when you pull client count from the controllers at certain intervals? I would at least verify that at x given time that the controller showed x number of associated client and PI showed x amount. PI will poll and depends how often you polling interval is. I don’t thing it matters what aps Or controller you have as long as PI is on a code that supports the devices you have. How does the chart look when you have it set for 1 month? Also if you have a guest network, can it be folks walking past? Don’t really know your environment is the thing.

[jmhouse96]

I do see fluctuations but have not been able to compare the two numbers. I do see clients lose wifi from time to time. Today there were about 8 people in a conference room each with a device connected to the wireless network and 3 of the devices all were dropped from the network while the others did not. After about 30 seconds those clients that lost connectivity were back online.

[jmhouse96]

Well TAC saw a client lose their IP address and advised this was due to the virtual IP of 1.1.1.1 and I need to get that updated. I did perform an upgrade and reboot all the controllers this weekend which seems to have helped. 

I will get the 1.1.1.1 updated when I can get DNS resources scheduled at a time I can reboot the controllers again.