• Gateway is controlled by a call agent e.g. CUCM
• Unique features
  • Centralized provisioning
  • QSIG supplementary services
    • Allow the passing of features transparently between PBX?
  • Centralized CDR (DS0 granularity)
  • MLPP
    • Boot someone off for a higher priority call
  • Hookflash transfer with CUCM

• MGCP identifier can addresses down to the individual port

• D-channel call-setup signals need to be carried in their raw form back to the Unified CM to be processed.
• Gateway terminates the data link layer and passes the rest of signals (Q.931 and above) to Unified CM via TCP port 2428
• D-channel will be down unless it can communicate with Unified CM

• The default selection order is Bottom Up which starts from channel 23.

• Cisco recommended way of configuring MGCP
• The gateway download the XML config file and parses it to convert to IOS commands
• When a changes is made on the CUCM, it will notify the MGCP gateway to download the config

• Add an MGCP gateway

• Configure the MGCP gateway
• The domain name must matches the hostname of the MGCP gateway including any domain names
• The product specific detail is only used if we are using MGCP auto configuration

• Add MGCP endpoints by selecting voice modules and voice interface cards (VIC)
• Use show inventory on the gateways to figure out the type and position of VIC

MGCPRouter# show inventory

NAME: "chassis", DESCR: "2801 chassis"
PID: CISCO2801         , VID: V05 , SN: FHK1244F23Y

NAME: "WIC/VIC/HWIC 1", DESCR: "WAN Interface Card - HWIC Serial 2T"
PID: HWIC-2T           , VID: V02 , SN: FOC14324S4N

NAME: "WIC/VIC/HWIC 3", DESCR: "VWIC2-1MFT-T1/E1 - 1-Port RJ-48 Multiflex Trunk - T1/E1"
PID: VWIC2-1MFT-T1/E1  , VID: V01 , SN: FOC124101BJ

NAME: "VIC 0", DESCR: "MFT-EC-32 - 32 Channel Hardware Echo Canceller"
PID: EC-MFT-32         , VID: V01 , SN: FOC12302R0G

NAME: "PVDM 0", DESCR: "PVDMII DSP SIMM with Two DSPs"
PID: PVDM2-32          , VID: V01 , SN: FOC1431436B

NAME: "PVDM 1", DESCR: "PVDMII DSP SIMM with one DSP"
PID: PVDM2-16          , VID: V01 , SN: FOC10443V5U

ccie-srst-gw#

MGCPRouter#

• Configure MGCP endpoints

• Configure T1 / E1
• Configure hostname and domain name

hostname mgcp-gw
ip domain name iptel.in.telstra.com.au

• Specify IP address of CUCM TFTP server hosting the configuration XML file
• Enable configuration download

ccm-manager config server <CUCM_TFTP_IP>
ccm-manager config

• Optional: Modify automatically applied MGCP configuration

• Configure the T1/E1 controller to be controlled by MGCP

controller T1 0/1/0
  pri-group time-slots 1-10, 24 service mgcp

• Configure q931 backhaul to call manager

interface Serial0/1/0:23
  isdn bind-l3 ccm-manager

• To enable the gateway to communicate with Cisco CallManager through the Media Gateway Control Protocol (MGCP
• Specify primary and redundant call agent

ccm-manager mgcp
ccm-manager redundant-host <cucm-pub>
mgcp call-agent <cucm-sub>

• [Optional] Configure other MGCP parameters

ccm-manager switchback immediate
mgcp bind control source-interface lo0
mgcp bind media source-interface lo0
mgcp dtmf-relay voip codec all mode out-of-band 

• Enable MGCP

mgcp

• Enable status poll - The advanced Service Parameter Change B-Channel Maintenance Status allows you to busy out unavailable b-channels on up to five MGCP voice ports. In order for this service parameter to work properly, you must enable status polling
• Inhibit restarts at PRI initialization - To allow restarts, uncheck this box. a RESTART or SERVICE message confirms the status of the ports on a PRI span. If RESTART or SERVICE messages are not sent, CUCM assume that the ports are in service.

• Advanced Service Parameters - In a production environment, you should always busy out unused b-channels on fractional circuits.

• Debug MGCP

Router# debug mgcp packets

Sep 25 08:00:48.247: MGCP Packet sent to 10.80.101.72:2427--->
RSIP 855851633 *@ccie-srst-gw.iptel.in.telstra.com.au MGCP 0.1
RM: restart
<---
Sep 25 08:00:48.251: MGCP Packet received from 10.80.101.72:2427--->
200 855851633 
<---
Sep 25 08:00:48.255: MGCP Packet received from 10.80.101.72:2427--->
AUEP 24 S0/SU3/DS1-0/1@ccie-srst-gw.iptel.in.telstra.com.au MGCP 0.1
F: X, A, I
<---

...
...

Sep 25 08:00:48.323: MGCP Packet sent to 10.80.101.72:2427--->
NTFY 855851635 *@ccie-srst-gw.iptel.in.telstra.com.au MGCP 0.1
X: 0
O: 
<---
Sep 25 08:00:48.327: MGCP Packet received from 10.80.101.72:2427--->
200 855851635 
<---
Sep 25 08:00:50.258: MGCP Packet sent to 10.80.101.72:2427--->
RSIP 855851632 *@ccie-srst-gw.iptel.in.telstra.com.au MGCP 0.1
RM: restart
<---
Sep 25 08:00:50.258: MGCP Packet received from 10.80.101.72:2427--->
200 855851632 
<---

...
...
Router# 

• L3 backhaul stuck in TEI. Seems to happen when we bind the source to loopback0

I've seen this is a common problem of everyone.
There are other methods but this method seems to be efficient enough to fix it without rebooting the router.

Assumptions; you'r ccm is properly bounded to loopback0 [confirm it using show mgcp command], your config are pointed to ccm properly.

Problem;
---------
show isdn status layer3 doesn't come up[multiple frames established], instead State = TEI_ASSIGNED

summary;
config t
int loopback0
shut
controller t1 0/2/0
shut

no mgcp

then once all those 3 are shut, no shut them one by one;
e.g.
int loop0
no shut
controller t1 0/2/0
no shut

conf t
mgcp