Bugs: CSCtf34858 - Severity 1 - catastrophic (WLC Code Levels: 6.0.182.0, 6.0.188.0, 6.0.196.0)

  

Catastrophic isn't my words, but Cisco's.  Engineer's beware ...

Client can't transmit traffic if it reassociates to an AP within 20 sec

Symptom: Intermittently, when a client reassociates to an access point (within 20 seconds after having roamed away from that same AP), the AP will fail to forward any data received from that client, to the wired network. Other clients can associate to that same SSID on the same AP and work fine. If the SSID is configured to use encryption, then, at the time the problem is occurring, decrypt failures will be seen, when the following debugs are enabled on the affected radio: ap#no debug dot11 dot11radio0 print printf ap#debug dot11 dot11radio0 trace print clients keys Workaround: Resetting the AP radio, or rebooting the AP, can cause the problem temporarily to go away. This problem is not seen in 5.2.193.0 or earlier code, nor does it affect autonomous IOS APs. Base Code: 6.0.182.0, 6.0.188.0, 6.0.196.0 Special Build: Following options are available: 1.     Move to 7.0.98.0 Release posted on CCO. Please note, 7.0 is a new feature release. 2.     Contact TAC to get a 6.0 Special or Beta release with fixes for the bugs below.  3.     Wait for the CCO release of 6.0 MR3 (Maintenance Release), which is planned for July/August 2010  The code is designed for 2100 / 4400 / 5500 / WiSM / WLC3750 / WLCM This Software Advisory Notice is issued against all the above Wireless LAN Controller software versions due to the following bugs:

Comments

[Dan L.]

Yeah, this was a fun one to troubleshoot until they announced it. Advisory was in Netpro discussion forum NOT in regular release note section. Good news is it seems to be intermittent in my case. However, when client is down, it stays down. Looks like some after hours firmware upgrade for me.

Good post!

[Stan]

Cisco ... how can you release this as safe harbor! BIG EGG ON FACE ... 6.x is the new 5.0 !!

[Nate237]

Be wary of the 6.x engineering code as a workaround... We installed it on our test site controllers and the kernel panicked within 10 minutes of booting.