AP disassociate from WLC

[denverelway]

We are currently using 2 Cisco WLC 5508 (software version 8.2.166.0), since early January we are experiencing many problems with one of our WLC. AP disassociate themselves from the controller for a few minutes and reassociate again. It usually starts in the morning when staff and students start to arrive and the problem continues until the end of class. 

In the evening when the number of customers decreases, there is no problem. The first days we observed the problem, we restart the controllers and the problem disappeared for a few days, but since Monday even after restarting the controller the problem persists. So we considered transferring the AP to our 2nd controller which initially worked very well, but following the transfer of some AP the problem appeared also between the AP and this controller.

There was no configuration change on the controllers between December and January. There are currently 144 APs on the first controller that has failed and 116 APs on the second. The APs found on the controllers are mostly 1142 and 2702. In the controller log files, there is nothing that can indicate a problem. On APs whenever the problem occurs, this kind of message can be found in the log files:% DTLS-5-SEND_ALERT: Send FATAL: Close notify Alert

We created a new port-channel between the controller and the Cisco 3750 switch, a network specialist came to analyze the problem and saw no apparent problem. Unfortunately, we have no support contract for these WLCs. 

Maybe someone has an idea?

Thanks

[ajc]

What I have seen when you get that error message is the AP not compatible with the Country Code configured in the WLC.

Check this link:

https://community.cisco.com/t5/other-wireless-mobility-subjects/wlc5508-ap2702-quot-dtls-5-send-alert-send-fatal-close-notify/td-p/3067057

[ajc]

2nd option, check the time on your WLC as indicated here:

https://community.cisco.com/t5/wireless-and-mobility/ap-can-t-join-dtls-connection-closed-by-controller/td-p/1871401

[Leo Laohoo]

---

@denverelway wrote:

software version 8.2.166.0 

---

This is helpful.  

Are the APs "jumping" in and out 2700/3700?  If they are, continue reading. 

When the APs jump back in, do they display have any of the following behaviours: 

1.  In the WLC, they show up with default AP name; or

2.  In the WLC, they show up with an IP address of "0.0.0.0"; or

3.  In the WLC, they show up with a status of "Download" forever; or

4.  Remote into the AP and look at the command "sh controller dot110" or "sh controller dot111" and the MAC address of the radio is "0000.0000.0000"; 

5.  If they have AP Groups configured, drill down to the AP and the AP Name, Location, and Primary/Secondary controller details are empty (or in default state) BUT the AP is assigned to the correct AP Group.

If the AP(s) has any of these, then I'd recommend to upgrade the firmware of the controller.  This has behaviour has been present since 8.0.X.X and wasn't fixed until mid-2018.  

One way to determine is to remote into the AP and look at the logs.  Before the AP jumps out, the logs will say that 5.0 Ghz radio will silently crash.

Kindly refer to Understanding Various AP-IOS Flash Corruption Issues.