This problem has me flummoxed.
Our infrastructure is larger than this, but for the purposes of resolving the issue, we have two sites with three Exchange 2013 MB/UM/CAS servers each (two DAGs of two active members in each site, with the standby third in the remote site). All Exchange 2013 servers are on CU8. Both sites use CUCM and Lync (I'm pretty sure Lync isn't part of the problem, but it is part of the environment). We have had Site A's CUCM-Ex2013 UM integration running for more than a year, but are just now getting it running in Site B.
We also still have Exchange 2010 installed. Getting CUCM-Ex2013 running is part of the prerequisite for migrating off Exchange 2010. CUCM-Ex2010 integration works in both sites.
For Ex2013, CUCM is configured with one SIP trunk for call routing, referenced by IP to port 5060 of each of Site B's active Ex2013 servers-- Servers B1 and B2-- and two SIP trunks to each of Server B1 and B2 for UM workers, referenced by FQDN to ports 5062-5068. SIP route patterns have also been configured for Servers B1 and B2, to handle SIP redirects from the Exchange Call Router.
Exchange UM is configured with several dial plans, one of which is for CUCM. A pilot number-- 2533-- is configured in the dial plan. For testing, I created a second pilot number-- 2532-- in the same dial plan.
The CUCM route pattern for 2533 currently points at the Exchange 2010 UM server, and that system is fully functional. 2532 points at the Exchange 2013 call routing trunk.
With Exchange logging set to high for UMCore, UMWorkerProcess, and UMCallRouter, I see the following events when calling 2532:
Event ID: 1650
The Microsoft Exchange Unified Messaging Call Router received a call with the following parameters:
Calling Party: "sip:+[calling number]@[CUCM IP]",
Called Party: "sip:2532@[Server B2 IP]",
Diversion Information: "",
Referred-By Header: "",
Call ID "ad3cf500-65511a3b-e8b-1141050a@[CUCM IP]",
Remote Peer "[CUCM IP]".
Event ID: 1622
A call received with the following parameters:
Calling party: "sip:+[calling number]@[CUCM IP]",
Called party: "sip:2532@[Server B2 IP]",
Diversion information: "",
Referred-By Header: "",
Call ID "ad3cf500-65511a3b-e8b-1141050a@[CUCM IP]"
was succesfully routed to "sip:2532@[Server A2 FQDN]:5062;transport=Tcp".
As you can see in the second event, instead of routing the call to a server in Site B, it is instead being routed to a server in Site A. Both Exchange call routers do this, and they also do it when I reconfigure CUCM to send the existing 2533 DN to Exchange 2013. When 2532 is directed at Exchange 2010, calling works.
Finally, when I point 2532 directly at one of the Exchange 2013 worker trunks, calling works: Exchange 2013 UM on that server answers the call and plays the welcome message.
So something screwy is going on with the call routing in Exchange 2013, and I am at a loss as to why.