I wonder if anyone more have seen this strange behaivure:
The AP's seems to work fine.
If you look in the WEB Page or in the CLI all seems fine.
Nearby wireless clients are not associated to the AP, but to other APs that are further away. The client can not get in contact with the affected AP.
Reboot the AP
The biggest issue is that the problem can not be recognized from error management within the AP and you can not see anything unnormal in the AP.
IOS ver 12.3.8JEA
Does anyone know if this cavet is resolved by Cisco in FW 12.3.8JEB
Of my knowlage, this have heappend on 10 of 3000 APs, but it is really affecting wireless industrial solutions. It is hard to discover, so it may have affected even more APs.
Using multiple SSIDs and different encryption methods, both WEP, dynamic WEP and WPA Enterprice.
Interesting... I have the same issue with clients that try to associate to APs on other floors of the building. No security is enabled at this level but users authenticate to BlueSocket/Active Directory.
What I did was enable Telnet and capture the "show tech-support" info and submit it to the OUTPUT INTERPRETER for analysis. What it is showing me is a large number of dropped packets due to shortages of numerous buffers. There are over 100 APs configured to channels 1, 6, or 11 which could contribute to these channels being overloaded. I also found that an Aruba test AP was recently installed in this area but, I have no access to check those settings -- could be some configuration/channel conflicts there as well.
Intresting fact that you have seen a large amount of dropped packets. What is the ratio between the dropped and total sent?
I have done som analyzes of the wireless traffic, and in my case, there is no problems with channel inerference or similar. I have used Airmagnet to do the analyze, and we use only ch 1, 6 and 11 as well.
In your case it could be a combination of channel overload / AP-overload. Check the amount of clients on each AP. It should not be above 20 and prefferable under 15.
BR / Johan
As it turns out... the Aruba test AP and sensor were responsible for the wireless client roaming from one AP/sensor to the other. I have performed site surveys using the AirMagnet software in other scenarios but, I find that it doesn't show me the directly connected client-to-AP information nor does it simulate what the user is experiencing. In my experience with the AirMagnet you have to watch the signal strength to see which device you are connecting to from a list of APs/devices.
So, I decided to perform the site survey with the "client utility" to see exactly which AP/device the client was directly attaching to and to get an idea of the signal strength variations. As it turns out every few seconds it was trying to change associations to another AP -- including the Aruba sensor.
In any case, since we have been testing the Aruba sensor and AP near the affected area, we disconnected the Aruba AP and sensor and performed another client test. The client stayed associated with the nearest Cisco AP with 54Mbps signal during this site survey when disabling and re-enabling the wireless card in several locations throughout the room. Then, we added the Aruba AP but not the sensor and tested the client again. When sitting at one end of the conference table near the open door, the client would assocate to the Aruba AP. Otherwise, the client always associated to the nearest Cisco AP. We left the Aruba devices disconnected to avoid issues until the problem is resolved. It's an old school practice but, it worked!
We still need to try to understand why it is that the clients were associating to the sensor and resolve that problem prior to implenting that technology in a mixed environment. Any suggestions about why the clients were connecting to sensors and APs in a roaming manner while stationary?
YES. As a new employee, I noticed that the WLAN has all APs in this building are only set to channels 1, 6 and 11 which certainly overloads the APs with clients in heavyly used areas such as the library and classrooms.
Thanks for your input/suggestions,
Please keep the APs on those channels. It is better to have two APs on the same channel nerby eachother compared to have them on nearby channels. Yes, you can do some advanced channelplanning by using the 1-4-8-11, or in europe 1-5-9-13, but be ware, it can end up to be even worce.
I do not have any experiance of the Aruba sensors. I know that the Client NIC now and then sends a NULL frame to get the time to scan for better SNR from other APs. It seems like the sensor is sending some kind of beacon into the air, or an dissaccossiate frame. Have you seen anything like it in Airmagnet capturing the frames?
The only solution to the issue of overloaded APs (to many clients) is to deploy even more APs. Use a decreased output power to create so called microcells.
Hi Johan - any luck with your issue? I'm having the same problem... I tried the latest 12.3.8.JEB1 but still the same problem.
For us, this problem is affecting about 10-12 APs of the 70 APs we have rolled out... we were going around to each AP trying to manually connect and discovered all these problems - hope there's a centrally manage software out there to detect the problem from remote..
If I come across any fixes, I'll let you know.
By the way - what model of APs are you using?
Thanks for your reply.
This means that there are more ppl with the same problem. I had some thoughts of a configuration error or similar, but then it would probably affect even more APs. Then I know that the JEB1 will not solve this.
We are using the 1100 and 1200 series, and I have seen this behavoiur on both types.
I hope to find a solution as soon as possible, because it makes the wireless enviroment a little unstable and unpredictive.
BR / Johan
Are you using VoIP over in this wireless area by any chance? If so, there are known issues with some devices and IOS bugs. This was something we experenced at my previous job.
No, we are not using any Vo-Fi devices in these wireless networks. We have one person in our company that beleaves that the AP Firmware is suffering from a memory leak, but then I think that is would affect more APs then it has today.
Today I called in to the support company to get some help from Cisco in this matter. This problem has now occurred several times and on even more APs. I hope for a fast answer to this error.