Exchange 2010 LTM v11 - SCCM AutoDiscover testing fails

Hello,

 

 

Here's a strange one:

 

My Exchange admin is using an SCCM script to test functionality of AutoDiscover on a CAS server. The server is trying to test itself. In other words, the server tests AutoDiscover by making a request of the LTM virtual server for the pool it's a member of.

 

 

In my European data center, this works. Exchange 2010, 2 CAS servers in the pool, LTM v10x

 

In my North American data center, it doesn't. Exchange 2010, 4 CAS servers in the pool, LTM v11.1.0 build 2027

 

 

I know that on my LTM v10 build, the load balancing algorithm randomizes the test accuracy because you can't guarantee that the test will be sent back to the server it came from. That fact aside, the test itself is successful, and SCCM is happy. (For this reason I will be writing an iRule to force tests generated by a server to be returned to the same server - This should be irrelevant to this conversation, however)

 

 

In the US build, the SCCM test fails with the error: (Generated on the CAS server)

 

VERBOSE: [14:04:25.776 GMT] Test-OutlookConnectivity : When contacting received the error Unable to connect to the remote server

 

VERBOSE: [14:04:25.776 GMT] Test-OutlookConnectivity : When contacting https://AD.network.dom/Autodiscover/Autodiscover.xml received the error A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because a connected host has failed to respond 1.1.1.1:443

 

VERBOSE: [14:06:25.790 GMT] Test-OutlookConnectivity : Scenario timed out. Scenario = 'Autodiscover: Web service request.', Overtime = '00:02:00'.

 

 

When he sets the hosts file to send that traffic directly back to it's own IP address instead of the LTM VS, the testing is successful.

 

 

The LTM v11 build is exactly per the Exchange 2010 manual deployment guide (We tried the iApp, but that was a... mess)

 

Everything is working just fine from a client perspective. The health checks are happy, and all is good. Except for SCCM testing.

 

 

It's almost like the LTM doesn't like sending requests to a pool when generated by a member of that pool.

 

 

Any ideas?

 

I can certainly provide more information, if necessary.

 

 

Thanks!