Constantly getting pulled in to fix either customer or wired network contractor Cisco installs. Usually changing controller settings to stop auto channel, limiting power adjust settings and limit the control the automated routines can do. But I have a number of customer sites where I just can’t get the clients to not randomly drop even though we have -65dbm RSSi and power at max of 16dbm of combined AP & antenna gain on 5GHz. Roaming isn’t occurring correctly or quickly even though we have client roam threshold set at -70. But I can’t stop the “sticky AP” syndrome.
Sorry I can’t be of any help, but to add seeing very similar symptoms on our installation mainly 3700’s Series APs and 9800 WLC (but migrated from 5508s which were experiencing the same issues). I’m interested if you find a resolution and I will post if I find anything concrete.
I do see the same issues. Also with Cisco you need to design for RRM if they are going to use it. APs too close to each other (RRM turns the power down so low) and clients with too many choices of AP to connect to. Remember the client makes the decision when to roam not the AP. Does the client have any roaming parameters . I have found turning of band steering to be beneficial if used and running separate SSID for each frequency.
Yeah, RRM mode, term stolen from Psion but means something really different, isn’t beneficial in a warehouse. You are viewed as a demon trying to get a customer to turn it off. When they called it Auto RF I had to get the customer to shut it all off and statically set the AP power & channels. They then complained that they paid $15K for a central console for the AP’s. Sorry, you didn’t ask. I have got the customers to limit the adjustment range in RRM so that we keep the DBM level within acceptable limits of both high & Low. Still get issues. Monitoring Mode causes many clients to go crazy when the AP drops them without any apparent reason to the point of having to reboot it. We have statically set channels on 2.4 but the 5.8 still drops even though we have not repeated any channels??? Next step is a WIFI monitoring software to see an entire net trace of the transactions.
We had a lots of disconnects from Cisco 2800 and 9130 series WAPs when used with the latest Intel chipsets. The problem revolved around 802.11r fast roaming/transition, which had been poorly implemented by Intel. Although it took Intel a VERY long time to resolve the issue (including their own purchase of Cisco WAPs), the latest Intel drivers are proving to be a lot more reliable.
The stable Cisco sites that we run, are those where almost every WLC auto-mode is disabled, and channel allocations are assigned manually. Cisco algorithm designers could benefit from spending some time in the heart of Wi-Fi dense cities, where channel hopping to find “clean air” is a non-productive interruption that must be avoided.
Hope this helps
Thanks Darren. this is good info. I too hate Cisco’s Auto routines, they are based on wireless office parameters and I do warehouses and manufacturing facilities. Clean Air is trash, Auto RF or RRM mode is more than useless when the AP’s have clear line of sight above warehouse racking but NO measurement from the floor where the client resides. There’s a big difference between 8-10’ ceiling height and no roaming and 45’ ceiling height with nothing but roaming, But Cisco has no clue.