Q: iPhone Network Stack Crash

I have been debugging an issue related to iPhones in our enterprise environment. I have hit some walls and wanted to request a little help.

Details:

- iPhone issue is intermittent effecting a relatively small sub-set of our users. (~25 of 300 users)

Description:

- at random times users report iPhone not receiving email communivations via Activesync, upon troubleshooting we found the entire network stack is frozen requiring the user to reset all network settings to resolve issue. (verified network stack failure by turning wi-fi off, connecting to LTE, and or connecting to other wi-fi networks, resulted in failure to retrieve data while in a continued failed state)

Commonalities:

- all devices have MDM profiles providing three payloads: Activsync (2010), Wi-Fi, General iPhone Settings

- all devices connect to 802.x enterprise wifi (Aruba controllers)

-  all users are located on particular floors of physical properties

-  once user experiences problem it becomes recurring for that user

-  all started on and after a particular date in January of 2015

Testing:

- restored iPhone resulting in continued failure of network stack

-  removed Airwatch profiles and manually added Wi-Fi and Activesync credentials resulting in continued failure of network stack

-  resored iPhone and manually added Wi-Fi network only and seeing no ill network effects

Other Infrastructure Notes:

-  upgraded Firmware and controllers for Aruba AP begining of December 2014

-  bypassed MS UAG for Activesync traffic (implemented change 11 days before iPhone issues appeared)

-  test wi-fi signal levels in effected areas and observed regular wi-fi signal drops and significant interferance

Questions:

-  I am leaning toward an issue with the way Activesync communicates to our load balancers and CAS servers after bypassing MS UAG. Is it possible that a misconfigured route a packet will take from Hardware load balancers to CAS can corrupt iPhone network stack?

-  Given that we can see issues with Wi-Fi signal strength is it possible that it can be currupting the network stack when Activesync initiates connection?

Next Steps:

-  Developed test groups with a variety of configurations (testing ongoing)

Recommendations for additional next steps or thoughts?