Gateway Endpoints Get Wrong IP Address

Unanswered Question
Apr 12th, 2007

I have an odd problem with the MGCP gateway at one of the sites in our recent 12 site deployment. Every few weeks we will get a report that inbound PSTN calls to this location ring once and then the caller gets dialtone (all inbound PSTN calls come in over loop-start FXO ports). We have discovered that if you go into CallManager and look at the gateway endpoints, they have taken on the IP address of the local WAN public address. If we do a "no mgcp" then a "mgcp" on the router, the address reverts to the ip address of the voice vlan, and all is good again, at least for a few weeks. This seems to be the only site that this is happening with. Our LAN runs over MPLS, which terminates at the router in a WIC-1DSU-T1. My personal feeling is that this may be a IOS bug, but I have not been able to find anything on Bugtracker or TAC Case Collection. I even had a TAC case open on it once, and all the engineer had me do was rebind MGCP to the voice vlan interface, which did not solve the problem long-term. I'm just hoping some of you out there may have seen this before and might have a solution. I have attached a copy of the router's sh version and sh run. You can see the WAN address under the serial0/0/0 interface. Can anyone shed some light? This is driving me insane. Thanks.

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
dgahm Thu, 04/12/2007 - 16:29


It could be a bug, especially since you are running 12.4T. Do you need the T code features? If not, I would suggest dropping back to 12.4 mainline. Another suggestion would be to use the loopback for your MGCP source bind. We always use the loopback and have never seen this issue with 12.3T or 12.4 code.

Please rate helpful posts.


allan.thomas Thu, 06/21/2007 - 02:10

Did this turn out to be a known issue? The reason being is that I have come across exactly the same problem on 12.4(9)T2.

Initially MGCP was bound to the loopback interface without having to bind it specifically. However after a temporary routing problem, which caused the ISDN backup to connect MGCP bound to this interface instead. Consequently when the ISDN went down MGCP was still bound to this interface, and therefore all incoming/outgoing calls failed.

The issue is still persistent even when you specifically bind the interface to the voice vlan sub-if. In this particular instance there was a power cut, and after the reboot MGCP bound to the loopback instead of the specific MGCP command.




This Discussion