Users are being randomly disconnected from the wireless APs, all users are not disconnected at the same time.
When Windows clients are disconnected they receive the "globe" rather connected icon. After about 10-15 seconds the connectivity is restored without the user taking any action.
There is no indication they are being disconnected from the wireless network. We have received another incident where a Apple Macbook was disconnected so we are certain this is not specifically a windows problem.
The SSID is configured for both 2.4Ghz and 5Ghz bands with PSK and the clients have a DHCP lease of 1 day gathered from a separate DHCP server.
WLC 9800-CL (16.12.4a)
10 x C9120AXI-E (220.127.116.11)
Disconnect occurs across the entire office space, do not believe this to be a specific AP issue.
As the time they are disconnected is so short, I initially wanted to go back to the users and advise that this just one of the issues of wireless networks in a shared office space. The client however is adamant they want to know why this is occurring.
All APs have been updated to attempt to resolve the issue as have the clients wireless drivers.
Copy of the radioactive trace logs attached when a disconnect occurred. I was unable to find an official reference for what the state transitions mean so I'm guessing that if a client was in "S_CO_RUN" it has previously successfully associated so when a client goes from "S_CO_RUN" to "S_CO_ASSOCIATING" it is an indication of a disconnect?
Also attached logs from a Windows client but they do not match with the actual times the client is disconnected.
Any suggestions on next steps to troubleshoot?
This is more investigation on each level : is this DNAC environment with ISE ?
start with below steps :
1. start listing what end device this occurs, or any devices ?
2. Only laptops or iPhones, all devices ?
3. what model of Wiless nic and Drivers on end user side.
Thanks for Reply. Will review other post.
No ISE although all APs are running lightweight to the WLC. SSID is configured for local breakout, not centralised switching.
So far confirmed issues have been identified on the originally listed HP laptop with Intel wireless and a single Apple (Intel) Macbook.
As the disconnect is so brief, it is not being noticed on mobile devices. For laptops it's more problematic as it can interrupt presentations and VoIP calls.
Users are not very cooperative, you know the age old "It works fine at home with no issue" response.
BB, config attached.
From what I understand Fast Roaming/Fast Transition (802.11r) are the same. Not enabled as yet but planning to change next week.
Not sure how this will help though as users are getting disconnected even when stationary so there is "good" reason they should be transitioning to another AP. All APs joins are stable for over 2 weeks.
- Also try to perform a client debug on the controller. you can have it analyzed afterwards with : https://cway.cisco.com/tools/WirelessDebugAnalyzer/
Thanks Marce, client debug I do believe has been replaced by Radioactive tracing on the newer controllers:
There is no "debug client" command any more:
- Ok. these threads are somewhat similar - you may find them informational :
Reviewing the suggestions, I've found that:
Will try enabling these next.
The most weird is that only Mac affected between all Windows clients that could ruin all theories, but let me introduce you to my last issue. Something similar happened to me while using 17.3.1 code. Windows clients were disconnected at the end of the session (session timeout configured under the profile).
Investigate a little bit on this and look if your clients are disconnecting after a period equal to session timeout which seems by default in your config. To see this, perform a packet capture in the client with Wireshark using the filter "dhcp". If you see many DHCP requests and no ACK until 180 seconds more or less, then you're hitting https://bst.cloudapps.cisco.com/bugsearch/bug/CSCvw30043/?rfs=iqvred . The bug seems to affect only 8.10.140 and 17.3.1 and not AP9100 series but who knows if you've discovered something.
Thanks so far for all the suggestions, the following changes were made today so we just waiting to see what happens:
Will feedback once more info available