10-20-2014 12:55 PM
Users have reported issues when running Mac OS 10.10 Yosemite when connected to our Aruba AP and WLC-based network, which uses WPA2 Enterprise PEAP authentication. In the logs, it appears that affected users are roaming between wireless access points every 5 to 20 seconds. It is unclear if the network is dropping the clients, or the clients are purposely trying to associate with a different AP.
There have been other reports of similar behavior on the Airheads forum:
There is a similar report on Apple's support forums, using PEAP with a different vendor's equipment:
I'm wondering if others have experienced this problem, stumbled upon a solution, or perhaps at least confirmed that this is a problem with Yosemite and not the network. Thanks in advance!
10-21-2014 01:31 AM
We're having exactly the same problem here. We had one user say when they used a different user id that the problem went away. We've also asked a few users to try created a new wireless service as shown in the following post but they haven't got back to us to let us know if it worked.
Let us know if you find anything.
10-23-2014 04:46 PM
I too was getting reports of Yosemite issues. upgraded my macbook - and it appeared OK - but did see some oddness on one of our ssids (using PEAP)
debug logs for my client showed this:
Deauth from sta: 28:cf:e9:xx:xx:xx: AP 10.112.x.x-00:0b:86:xx:xx:xx-00-ap105-xxx Reason Tx fail reached maximum
noticed that for this particular ssid profile I had max-tx-fail set to 25
returning that value to 0 [default - unlimited fails allowed]
my macbook returned to its previously... more reliable state.
Let us know if your problem ssid profile also has a non-zero value and if changing that appears to help. I've only just made the change global in my environment so don't have any feedback - and the ssid impacted was for our students in the dorms.... feedback is not always a priority - and even less of a priority it seems when its positive :)
It was a setting I configured a while ago - with probe request threshold - to help mvoe sticky clients along - and prevent clients from making bad association decisions.... pre client-match
There's something about about allowing unlimited fails to get a device working.... I'm sure a source for much humor / life lessons - but lets get some feedback first
10-24-2014 01:24 PM
I had noticed the "Tx fail reached maximum" messages, but only related to APs from which the client had already roamed. In other words, it appears the the Tx fail is hitting the max threshold due to a station disconnect. Who knows, though. I'm hesistent to lower this value, as we need clients to roam when appropriate.
I have, however, had some success by disabling Bonjour on affected machines, but I don't think you would want to do something like that on a student's computer.
Given the diversity of vendors' equipment that has been affected by this problem, I probably won't be performing additional troubleshooting for now, as the issue seems to be attributable to OS X, and Apple will likely issue a patch. (We experienced similar issues with 10.7.)
10-25-2014 09:18 AM