Hi all, I recently implemented OpenSIPS to load balance calls to an asterisk cluster, using the dispatcher module and taking advantage of the active probing for failed destinations feature.
Everything works like a charm, except when one of the destinations goes to probing or inactive state. When this happens, all of the calls that would have been made to that node are sent to the next one. To be clear, this is not me using ds_next_dst() from FAILURE_ROUTE, this is once the node has already been put in probing or inactive. Also, the problem is not the hashing algorithm. I am hashing on call-id, and it gives me a very even spread between my servers. However, when one of them goes down, all of the calls that would have been allocated to it are instead allocated to the next server in the list. This creates one server with double the calls, overloading its capacity. It would seem simple to hash the calls only over the active servers. I am using version 1.6.2. Is this something that is fixed in later versions? Or am I not configuring OpenSIPS correctly? Regards, Alejandro _______________________________________________ Users mailing list [email protected] http://lists.opensips.org/cgi-bin/mailman/listinfo/users
