[OpenSIPS-Users] Load_Balancer module - disabling GW from routing script

[Alejandro Recarey]

Hi all,

I am currently trying to enable GW disabling from inside the routing
script with the load balancer.

I managed to get it working reading the module documentation, but now
I have a problem. I want the gateway to be disabled only if it is
unreachable, that is, if it does not answer at all to the SIP INVITE
being sent, but the Asterisk Gatways are being disabled all the time,
every 5 - 10 minutes. The part of routing script that takes care of
the load balancing is:


. . . (ommited) . . .

# ---- Outbound to Load Balancer---- #
    if(uri=~"^sip:5378[0-9]*+@"){
    route(7);
}


# ---- Load Balancer ---- #
route[7] {
    if (load_balance("2","pstn","1")) {
        if (is_method("INVITE")) {
                t_on_branch("2");
                t_on_reply("2");
                t_on_failure("2"); #custom failure route for LB
        }

        if (!t_relay()) {
                sl_reply_error();
        };
        exit;

    }
    else{
        sl_send_reply("503","Service full");
        exit;
    }
}


# Failure route for Load Balancer
failure_route[2] {
        if (t_was_cancelled()) {
                exit;
        }

        # If status is 408 - Timeout, disable route
        if (t_check_status("(408)")) {
                lb_disable();

                # Try to load balance once again
                if ( load_balance("2","pstn","1") ) {
                        t_on_failure("1");
                        t_relay();
                } else {
                        t_reply("503","Service Unavailable");
                }
        }
}

If I execute opensipctl fifo lb_list:

Destination:: sip:x.x.x.x id=5 group=2 enabled=yes auto-re=on
        Resource:: pstn max=90 load=13
Destination:: sip:x.x.x.x id=6 group=2 enabled=yes auto-re=on
        Resource:: pstn max=60 load=9
Destination:: sip:x.x.x.x id=7 group=2 enabled=yes auto-re=on
        Resource:: pstn max=60 load=9


As you can see, I only disable the GW if I receive a SIP 408 message.
I thought this message would only be generated if the SIP INVITE timed
out, but it seems Asterisk is sending a 408 message for normal call
reasons (maybe when the dialed party does not pick up and the dial
times out??)

This is a huge problem because my termination GW are being disabled
all of the time, and randomly. Since my OpenSIPS server is in
production and I cannot reload config until tonight I have setup a
cronjob that enables the gateways every minute, but it's hardly an
optimal solution ;)

Does anybody have experience with the setup described? (OpenSIPS
Load_Balancer + Asterisk with in script disable of bad GW's?). Any
pointers would be appreciated, thanks!