Hairpinned call failing

[mmelbourne]

We have a DPNSS (converted to QSIG via Westell boxes) link between our CallManager and Centrex voice network. Users are encountering problems when a Centrex extension is diverted to an IP Phone. If another Centrex extension calls the Centrex extension with the divert, the call if forwarded as expected. However, if another IP Phone calls the Centrex extension with the divert back to an IP Phone, then CallManager generates the "your call cannot be completed" message.

It appears that CallManager is preventing the call from hairpinning out of the IP network and back in again via the QSIG/DPNSS trunks. Workarounds so far, have been to add a translation pattern so the Centrex DN is translated to the DN of its forwarded destination which saves on trunk capacity, but assumes we know when such diverts have been put in place.

Is there anything which can be done to allow this to work (albeit with suboptimal routing)?

[agopala]

can you make sure that the original caller and the incoming GW both have a CSS that can search the partition in which the destination phone is located in..

[mmelbourne]

Yes, the CSSs are correct. In fact if I take the divert off the Centrex I can call the Centrex phone and manually transfer it back to an IP Phone. But it doesn't work with a permanent divert on the Centrex phone.

[agopala]

Ok, if you are sure the annunciator is giving the message, we should be able to find in detailed CCM traces why. I wonder if we are not receiving all the digits or something like that with a permenant divert

[mmelbourne]

From the traces, it turned out that the diverted number was being returned from the Centrex network with the Node ID (access code) prefixed onto the 4-digit number, although all our 4-digit extensions are unique. A simple translation pattern to remove its own Node ID prefixes fixed the problem.