I am having trouble figuring out why my outbound calling is not working in SRST. During normal operation calls are sent to the GW with 11 digits so I setup the SRST dial-peer to forward 11 digits on LD calls to the carrier. I have tried forwarding all, 10 and 7 but nothing works. When I ran debugs on the ISDN q931 it does not even show the call, it is like the dial-peer does not forward the call to the serial interface. Any ideas what I am missing here? I have attached the debugs of the dial-peer.
Here are the details:
All phones are SCCP.
MGCP controlled Gateway with PRI/T1
dial-peer voice 40000 pots
description ***SRST Long Distance Outbound***
incoming called-number .
no ip address
isdn switch-type primary-ni
isdn incoming-voice voice
isdn bind-l3 ccm-manager
isdn bchan-number-order ascending
no cdp enable
ISDN q931 does not show the call hitting the T1
Debug Dial-Peer shows call hits DP 4000 Pots
Match Rule=DP_MATCH_DEST; Called Number=719194400699
005280: Apr 10 15:33:08.549: //-1/D0C23845AE4C/DPM/dpMatchPeersCore:
Result=Success(0) after DP_MATCH_DEST
005281: Apr 10 15:33:08.549: //-1/D0C23845AE4C/DPM/dpMatchSafModulePlugin:
dialstring=719194400699, saf_enabled=1, saf_dndb_lookup=0, dp_result=0
005282: Apr 10 15:33:08.549: //-1/D0C23845AE4C/DPM/dpMatchPeersMoreArg:
List of Matched Outgoing Dial-peer(s):
1: Dial-peer Tag=40000
It looks to me like there is a loop on matching the 40000 dial peer in the full debug.
You are showing your SRST config. Can you also show the part of your config where you configure mgcp fallback? The section will start with "call-manager fallback-mgcp".
Next time you want to test like this, put an ACL on your remote gateway that blocks the IP addresses for CUCM inbound and outbound. That will cause the router as a whole to fail over.
Your underlying problem was not that the phone did not failover to SRST, because it did. The problem was that the MGCP process on the router did not failover. So the easiest thing is do a test failover for the entire remote site, perhaps during a maintenance window or over the weekend.
Block CUCM to the entire remote site:
access-list 100 deny ip any host <ip.of.cucm>
access-list 100 deny ip host <ip.of.cucm> any
access-list 100 permit ip any any
interface gi 0/0 (inbound WAN interface at the remote site)
ip access-group 100 in
ip access-group 100 out
(You'd think you would not need both directions for the ACL, but in my lab it's the only way to make it work.)
It is possible to block CUCM to just the phone, but to block it to the MGCP process requires that MGCP is bound to a specific IP on the remote router as is call-manager-fallback. Using a loopback for both is a good idea.
access-list 101 deny ip host <ip.of.phone> host <ip.of.cucm>
access-list 101 deny ip host <ip.of.cucm> host <ip.of.phone>
access-list 101 deny ip host <ip.of.loopback> host <ip.of.cucm>
access-list 101 deny ip host <ip.of.cucm> host <ip.of.loopback>
access-list 101 permit ip any any
A third possibility is to create a CMG with just one call processing node in it, create a new device pool with that CMG (and the SRST reference, etc.), assign that DP to the MGCP router and to the phone (which will require an apply-config on both), and then stop the Cisco CallManager Service on just that one CUCM server. This will cause both to failover to SRST.