4402 clients intermitting disconnecting

[walter001111111]

We have a setup with one 4402 WLC and 7 AIR-LAP1142N-E-K9 APs in an environment with around 200 clients.

Without any clear reason, clients are intermittent loosing internet connection.

Client are registered to be connected to the controller, however are unable to reply to ping, and from the client it is also fails to perform ping checks.

This happens using 802.11x authentication, but also when using WPA2 authentication.

Firmware on the controller is 7.0.240.0.

Are there any newer versions still available? Could upgrading firmware help in this situation?

Does anyone has other suggestions on solving this issue?

[Philip D'Ath]

Is it happening to every client at the same time, of randomly to individual clients?

There could be a million reasons if it is randomly happening to individual users - such as noise.

Have you considered upgrading to something newer than can support CleanAir, like 2702i access points (which would also need a new WLC)?

[walter001111111]

This happens to individual clients, however the APs with the highest number of clients seems to be impacted the most.

Roaming seems to be working correctly, disconnection happens mostly while clients are static.

Channel settings are fixed, power level is fixed, so changing channels or power level on an AP shouldn't be causing the problem.

When we can't fix the issue on the current setup, next logical step is to upgrade the equipment.

[Philip D'Ath]

Any chances the clients are getting disconnected while idle?  Could be a power saving issue.

[walter001111111]

When clients go idle, it is indeed expected clients are disconnected.

Might need to re-word, not disconnected, but not replying to ping anymore. But when awakening, the connection should be active and working again.

And loss of network activity is apart from ping checks (which where setup to make the problem visible) also detected from the client side.

[walter001111111]

Based on the documentation of the WLC, I was in expecting a max of 128 clients per AP.

The clients are not evenly spread over the building, causing some AP to have 4 clients, and other have max 45 clients.

Searching for more details on this, looks like Cisco has a rule of thumb of 25 clients per AP, all depending on the type of usage.

See eg https://supportforums.cisco.com/discussion/11320491/number-users-1142n-access-point

Guess this would mean either adding some more AP to the installation, or reshuffling the current over the building.

[Prabath Godevithanage]

glad you've got to the bottom of it,you've noticed that you could set a max number of clients per AP for WLAN,if you don't have enough coverage this will introduce more issues rather than solving your problem,perhaps you could use that for testing

given that you only have 7 APs to play with I would just add more to cover up your high user density areas rather than reshuffling

[Prabath Godevithanage]

Are there any newer versions still available?Yes. there are 3 more versions above 7.0.240

Could upgrading firmware help in this situation?if it was stable before,it may not but keeping in mind that the number of clients may have gone up,there might be new wireless cards,drivers etc..

Does anyone has other suggestions on solving this issue?are you able to narrow it down to a specific client models or wireless cards.I noticed you have mentioned that you are noticing it more when the clients are static,does that mean when the device is idle as in not in use or in use but not moving?

[walter001111111]

The issue happens on all clients which are used, like MacBook, iPhone, Samsung Android devices.

What was ment with "static" was that the client was not moving. Normally that should mean there would be no roaming action needed.

[Prabath Godevithanage]

have look at the "session time out" setting to make sure that no one has touch that. Also do you have areas in particular with no issues at all or is it through out the network,

I was trying to suggest you  few ways that you could narrow it down and replicate the issue to do some debugs,This seems very random. What's the possibility of setting up a new SSID with exact settings and lock it down to one your APs. Then you can setup a one of your spare laptop to see the behaviour.

[walter001111111]

Session timeout was set earlier to 28800 seconds. Has been updated to 43200.

And together with other changes, it looks like the situation has improved.

There where some areas in the building where almost no issues where noticed. Might also be caused because those areas are less crowded (less users), and using less bandwidth.