Cisco Support Community
Community Member

Client DNS over CAPWAP Question, take a look.

I was callled back into work Friday for an issue with massive slowness over wireless connections with our EMR software.  Not truely believing it was a 'wireless' issue when I came in and did some throughput tests, I immediately started looking at simple issues and found that I could not ping one of the EMR servers due to DNS fail.  Despite everyone else that was at work that day knowing that NS1 was down, but NS2 was fine (verified) no one correlated the potential slowness issue to DNS as a wired client worked fine.  In doing a sniff of the traffic on my laptop, you could easily see where the lookups were failing and then eventually would be tried via NS2, then back to NS1 and fail and to NS2.  (We use bluecat i.e. blackcat by the way).  However, when wired, you might see a NS1 fail but then NS2 was then used all the time and you wouldn't notice any slow downs obviously.   So, the question is, why would a wireless client not completely failover to NS2 - ALL THE TIME - like a wired client.  Is there some function of DNS that is controlled by the controller, I don't recall reading anything in regards to this, but you do have to supply both DNS servers when creating an interface.  On the other hand, is there a known issue with Win7 (maybe XP, verifying the types of clients) that only the wireless interface does not change NS servers for an extended period of time?

Thoughts? and thank you,



Client DNS over CAPWAP Question, take a look.


Good troubleshooting...

Just my 2 cents here...

The WLC acts as a layer 2 transport for wireless frames to the wired. I am not aware of ANY DNS shaping done by the WLC at all. I suspect its the WiFi client and what it is ultimately choosing. Have you tested more than one flavor WiFi client?

"Satisfaction does not come from knowing the solution, it comes from knowing why." - Rosalind Franklin
Community Member

Client DNS over CAPWAP Question, take a look.

At the time of slowness, the clients reporting would have been Intel and Ralink for sure, maybe broadcom.  Problem is, by the time I was called back in, it was end of day and all the clinics either had given up or were closed so no test users available other then me and the EMR guy and we had Intel 6205's, one of us using the Intel supplicant and one not.

By the way, to clarify further, this is through a Citrix Client, so you initially hit the main Citrix (web interface) server which creates an ica file to point you to an application server and the client talks to the application server. So you are not actually watching EMR traffic, but Citrix traffic from client to server and seeing the lookups fail wireless, but not wired.  So being a Citrix guy as well, I'm looking from that angle but I really doubt it, just like I doubt the controller having anything to do with it.  My gut/past experience tells me it goes back to the microsoft somehow, but looking everywhere.

CreatePlease to create content