cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
307
Views
0
Helpful
2
Replies

H.323 Failover for CCM doesnt work well..

seanwy
Level 1
Level 1

Hi folks,

had recently tested out a H.323-based voice gateway with 2 CCM (Publisher & Subscriber). As norm, we had the subscriber as the Active & Publisher as Standby for failover.

We then made a call out to the PTSN from an IP Phone, had the Subscriber CM services stopped, and expected the Publisher to failer over immeditaely w/o dropping the ongoing call. However, the call dropped & there was a message displayed on the IP phone indicating the CM has fallenback, and features were disabled. It then took about 20 to 30s for the IP phone to be reverted to the Publisher,but only after some push of the phone's button.

This wasnt the case when MGCP was used for the gateway.

Below is a brief description of my configurations. Would appreciate any feedbacks/comments. Btw, using CCM 3.1.4a, IP Phone firmware 4.8 & IOS

12.2.10B.

dial-peer voice 100 voip

preference 1

destination-pattern 6370....

voice-class h323 1

session target ipv4:10.228.204.5

dtmf-relay h245-signal h245-alphanumeric

codec g711ulaw

ip precedence 5

!

dial-peer voice 1 pots

destination-pattern 0T

progress_ind setup enable 3

progress_ind alert enable 8

progress_ind progress enable 8

direct-inward-dial

port 2/0:15

!

dial-peer voice 101 voip

preference 2

destination-pattern 6370....

session target ipv4:10.228.204.4

dtmf-relay h245-signal h245-alphanumeric

codec g711ulaw

ip precedence 5

!

Thank u!

2 Replies 2

Vance Krier
Level 4
Level 4

Hello,

I'm pretty sure that's expected behavior. In an h.323 call, the CM is actively involved in the call and acts as a 'middle man'. If the CM process fails, is stopped or otherwise becomes unavailable, the call drops as you've described. MGCP doesn't have that limitation because the gateway and phone can communicate directly without the intervention of CM once the call is up.

At least that's how I understand it.

HTH,

Vance

er hunh, then whats the point of failing over when the ongoing call will be disrupted despite claims being made that existing calls will not be dropped, unless MGCP is used..? Is there any way 2 smooth out this H.323 failover transition ?

Any advise ? Thanx!

regards,

seanwy.