Voice Gateway Failover Issue

ray.carter · ‎05-03-2010

90 trunks are distributed across three VG's andthe call flow should be as

> follows;

>

> Call 1 out via vg 1

> Call 2 out via vg 2

> Call 3 out via vg 3

> Call 4 out via vg 1

> And so on

>

> This distribution is acheived by setting up the route group on call

> manager as cyclic however we have found that if one isdn is down say

> on vg 2 the following error occurs

>

> Call 1 out via vg1

> Call 2 caller receives busy

> Call 3 out via vg 3

>

> Hence in this senario 1 in 3 calls will fail. The way it should work

> is that the call destined for vg 2 should reroute out of vg3,however

> it appears that the call manager is interpreting the sip messaging

> incorrectly and not rerouting the call to the next vg as designed

The gateways use SIP as the protocal that signals to the call manager

The call manager is running 7.1.3

I would be interested if anyone has had the same issue when working with this type of set up?

Tommer Catlin · ‎05-03-2010

I think there is a setting within the Service parameters to force it try the next available trunk. Its turned off by default. We had this problem with h323 gateways in a route group. I cant recall the parameter, but i can search around for it.

Also, make sure you are running the latest and greatest IOS for the gateways... there are many SIP fixes in there. But I think your issue is the service parameter setting.

I think this parameter: (taken from another post)

If your PRI is down then the router should report back to CCM a cause code 1 unallocated number. **Editing because I had the syntax wrong** By default, Stop Routing on unallocated number is set to True, which is the CCM equivalent of that cause code. You need to set this to false since the default stops routing. Assuming your CCM Route lists and groups are working (which I do recommend a reset or the new changes won't take effect) then it should failover to the alternate H323 gateway.

ray.carter · ‎05-04-2010

Thanks for the reply I have seen this parameter and set it to False with no effect unfortunately.The gateways concerned are connected via SIP trunks so I suspect the issue is in this area.

Eric Krieger · ‎05-05-2010

Hello Ray,

Please configure the following command in the global mode on all your vg's:

"no dial-peer outbound status-check pots"

This disables status checking of an outbound POTS dial-peer.

This way, the dial-peer is matched, even when the ISDN is down (and voice-port), and reports a cause code with a "No circuit/channel available" back to CUCM which then hunts for the next member (vg 3 in your case).

Without this setting, the dial-peer is down and therfore not matched, this results in a cause code with "Unallocated/unassigned number" sent back to the CUCM which stops hunting (giving you a busy).

Good luck!

Eric

ray.carter · ‎05-06-2010

Thanks for this Eric, I will let you know how I get on with this.

Ray

stephen.himebaugh · ‎05-07-2010

We had a similar problem with H.323 and CallManager 7.x. Maybe this will give you a clue as to how to fix with SIP

Calls were not failing over to second route group when only PRI was down on only (H.323) voice gateway in first route group. No other H.323 gateways or MGCP interfaces in first route group. Voice gateway is 12.4(24)T.

Cisco TAC had us do the following which fixed it:

For the voice gateway, input the following commands:

voice service voip -> h323 -> no h225 timeout keepalive

On CallManager, change following Service Parameters:

Stop Routing on User Busy Flag -> FALSE
Stop Routing on Unallocated Number -> FALSE

Tommer Catlin · ‎05-07-2010

I think that is the key... I remember doing this with multi-h323 gateways