Hold & Transfer not working at remote sites

[Asad Raza]

Hello everyone,

One of our clients is experiencing an issue with being able to perform a simple hold and transfer function at their remote sites. They currently have a CUCM 8.6 at their HQ, with two remote sites using the WAN for signaling, but PSTN for all external calls, even for intersite dialing. Their remote site gateways are running an AA tcl script. Users at the HQ are experiencing no problems with any fucntions however, if a user from the HQ dials a remote site, the AA script kicks in as expected, and the extension is dialed to reach a user. Once the appropriate extension is dialed and the call is connected to the user, the call proceeds normally. However, if the user at the remote site places the HQ user on hold, MOH is not streamed. Also, if the remote site user attempts to transfer the HQ user to another internal extension at the remote site, no music, nor any ringback is heard. The call is transferred, but audio becomes one-way only where the HQ user can hear the remote site user, but the not the other way around.

I would also like to point out that all gateways are registered as a SIP trunk to CUCM whereas all IP phones use SCCP. Could this be a problem? Also, I have configured the phones, MRGs/MRGLs/MoH/MTP at HQ and remote sites in exactly the same manner. However, I have registered a hardware MTP resource on the remote site gateway with the CUCM and assigned it to its device pool (Device pool for this remote site).

Can someone shed some light on what the issue may be? Any help would be greatly apprecaited. Thank you!

Update: I have attached a snapshot of the topology to better demonstrate the current network.

Regards,

Asad

[Robert Craig]

It's almost like the Remote users are invoking the hardware MTP when they use supplementary services (On hold, Transfer) and the MTP IP isn't allowed in a firewall somewhere or not even in the routing table.

[Asad Raza]

Hi Robert,

Thank you for your response. I don't know if I should have mentioned this, but all communication between the HQ and the remote offices takes place over a VPN connection through their global HQ in UK. Most likely they are running some kind of firewall, but I fail to understand how it would interfere in the operation since the MTP at the remote site is on the local gateway, and the local gateway is able to communicate with the CUCM. Also, the MTP shows up as registered on the CUCM.

Any other ideas?

Regards,

Asad

[Robert Craig]

It still seems like a routing issue. If I am interpreting the design correctly, all phones are registered to the CUCM at the HQ and you are using each sites voice gateway as a Local Route Group for external calls. Any other calls use the Pub for signaling. Thats just what I can tell. Regardless, media negotiation is done once the call is transferred. Once that happens, it's back to phone to phone traffic flow, unless they are using a MTP for some odd reason (bouncing through a sip trunk, transcoders locally, etc.). So, in the case of one way audio, the RTP stream from HQ to the Remote side is not reaching the phone or transcoder/mtp. It could be a firewall somewhere affecting RTP ports in one direction, or some kind of weird routing issue.

[Asad Raza]

Hi Robert,

It turns out that it was a codec mismatch problem. The incoming dial peer was statically configured to use a single codec. Once we created a voice class with all the commonly used codecs, everything started working just fine. The hold, transfer, and MoH problems all went away. In fact, it resolved another problem that that the client was experiencing as well, which was that whena remote phones were called from the PSTN, the voicemail did not engage. Once the codecs situtation at the dial peer was resolved, the voicemail started working too!

Thanks for all your help!

Regards,

Asad