We have a classic problem with Client not Authenticated but with a specific twist:

1. We have an Exchange 2010 cluster, let's say EX01 and EX02, the connection is always to smtp.acme.com, then it is switched through load balancer.
2. We have an application server, call it APP01 There are clients connected to the APP01.
3. There is a need for anonymous mail relay from both clients and APP01.
4. The Anonymous Users setting of the Exchange is DISABLED, but the specific computers - APP01 and clients by the mask, let's say, 192.168.2.* - are enabled. For internal relay, a "Send Connector" is created, and then the above IP addresses are added for the connector to allow computers, servers, or any other device such as a copy machine to use the exchange server to relay email to recipients.

The problem is that the relay works for APP01 and some clients, but not others (we get "Client not Authenticated") - all inside the same network and the same mask.

This is basically what we do to test it outside of our application: http://smtp25.blogspot.sk/2009/04/530-571-client-was-not-authenticated.html

So, I am looking for ideas:

• What can be the reason for such a strange behaviour?
• Where I can see the trace of what's going on at the Exchange side?