odd problem with dual integration of Unity 4.0(4)SR1

[jshapura]

We have a Unity 4.0(4)SR1 cluster doing unified messaging w/Exchange 2007 and integrated with our production Callmanager 3.3(3) cluster. This has been running smoothly for several years.

Recently we brought up a test Callmanager 4.2(3) cluster on new servers and as part of the testing set up a dual integration with our Unity cluster. The Unity cluster was licensed for 48 ports; as we were only using 32 for the prod CM 3.3 integration, we used the free 16 ports for the test CM 4.2 cluster. We followed the CM 4.2 Integration Guide for Unity 4.0 instructions, and all was working fine (both CM clusters could talk to Unity, ports were answering as expected (first 32 for production CM 3.3, last 16 for test CM 4.2)).

Fast forward to yesterday. One of our guys decides to change the test CM 4.2 voice mail port config. From the test Callmanager 4.2, he deletes the voicemail ports defined for CM 4.2 ONLY, doesn't touch any of the production definitions. Nothing is changed in Unity; the new integration in Unity was left in place. Soon we were getting calls from users of the production CM 3.3 phone system saying they could not play back messages from Outlook via VMO, though regular voice mail via phone was working. Looking at the event log in Unity we saw lots of the following message - 'An attempt to turn OFF the MWI for extension (prod extension) failed because Cisco Unity lost connection with Cisco Callmanager (test 4.2 system)'.

So we watch the Port Status Monitor and sure enough it looks like Unity is attempting to use the ports previously defined for the test 4.2 system for production MWI callbacks.

We immemdiately deleted the new test integration in Unity and all was well with production users again.

Has anyone run into this before? We don't understand why the 1st (production) integration started using ports that were previously dedicated to the 2nd (test) integration. Is it expected this would occur if the test integration ports get deleted on the CM side? Just want to make sure we understand this before re-implementing the test integration again.

Thanks in advance for your help!

[eschulz]

The behavior here depends on how you added the second CCM into UTIM. If you created a second cluster within the first integration, then Unity does not differentiate these as separate phone systems. Subscribers are only associated with the integration as a whole and Unity considers ports from both clusters to be valid. Thus outbound TRaP and Notification calls will look for the first available port (top-down search) within that integration. Since your test CCM was added second, it likely has all the higher port numbers. Therefore, the top-down search will result in Unity almost always trying to use these ports.

The expected scenario for having multiple CCM clusters within a single UTIM integration is that transparent dialing can occur between the clusters. i.e. you should be able to pick up a phone on cluster2 and dial over to cluster1 just as easily as if you started on cluster1.

In Unity 4.0(4) this was you only choice for configuring direct connections to multiple CCMs. However, with Unity 4.2(1) you could instead create multiple CCM integrations such that Unity treats each as a distinct phone system. This then allows subscribers to be associated with a specific CCM. TRaP and Notification calls for subscribers will then be routed out through ports on the associated CCM.

Hope this clears things up a bit.

-Eric

[jshapura]

The interesting thing is that before we deleted the test CM 4.2 ports on the Callmanager, the Unity ports WERE acting as if they were dedicated to each CM instance (i.e. watching the Port Status Monitor we would only see ports 32 and down used for MWI and trap by the production CM 3.3 extensions (as well as VM), and ports 33-48 were only used by the test CM 4.2 users. This did not change until we deleted the CM 4.2 ports on CM; then Unity was trying to use all 48 ports for the production CM 3.3 integration.

Now I'm trying to figure out if I really can do this with Unity 4.04SR1 without causing me grief on the production side. I understand your point about transparent dialing, and this is the case between my clusters (I can call from one side to the other), but I must be missing something else, as witnessed by the error log messages I referred to earlier (repeated instances saying MWI notification was failing for the production CM 3.3 users via the previously allocated MWI ports on Unity for CM 4.2 because the CM 4.2 integration no longer existed).

Is there any more detailed doc that discusses this other than the CM Integration Guide? Or am I safe as long as I ensure the integrations stay in place on both CMs and Unity)?

Thanks in advance for your advice!