[SR-Users] Via branch parameter in end2end ACK

[Francesco Castellano]

Dear sirs,

using version 3.1.3 (kamailio flavour), I noticed that the Via
inserted by the proxy, relaying an end2end ACK, contains a branch
parameter equal to 0 (with no magic cookie prefixed). No misbehaviour
followed, but I was wonder about the reason for this behaviour and
whether it can be a fault of my config:

route
{
    [some checks]

    if (!is_method("REGISTER|MESSAGE|PUBLISH")) {
        record_route();
    } else if (is_method("REGISTER")) {
        route(HANDLE_REGISTER);
        exit;
    }

    if (loose_route()){
        if (!has_totag()){
            if (!is_method("ACK")){
                send_reply("403", "Preset Route Set Not Allowed");
            }
            exit;
        }

        t_check_trans();

        if (is_method("ACK")){
            if ($du == $null){
                handle_ruri_alias();
            }

            if (isflagset(T_UAC_NATED) && isflagset(T_UAC_USER)){
                add_contact_alias();
            }
            remove_hf("Proxy-Authorization");
            t_relay();
            exit;
        }
    }

    [other processing]
}

Here's an example for the forwarded ACK:

U 2011/05/17 15:05:49.392201 proxy_1_ip:5060 -> location_callee_ip:5060
ACK sip:5303838 at 192.168.130.169;user=phone SIP/2.0.
Record-Route: <sip:proxy_1_ip;lr=on;ftag=a2z0myc9r6>.
Via: SIP/2.0/UDP proxy_1_ip;branch=0.
Via: SIP/2.0/UDP proxy_0_ip;rport=5060;branch=0.
Via: SIP/2.0/UDP
192.168.130.171:2048;received=location_caller_ip;branch=z9hG4bK-pud4cfiz8vsh;rport=4791.
From: "cento tondo" <sip:100 at proxy_0_hostname>;tag=a2z0myc9r6.
To: <sip:5375485 at proxy_1_hostname;user=phone>;tag=00221b79b1b28969.
Call-ID: 3c352c0b8419-9tbq023tzfkp.
CSeq: 2 ACK.
Max-Forwards: 68.
Contact: <sip:100 at 192.168.130.171:2048;alias=location_caller_ip~4791~1;line=gwwqwi25>;reg-id=1.
Content-Length: 0.
.

where proxy_1 does force_rport and record_route() on ACK too, and
proxy_0 does not. Both proxies run the same version, and a similar ACK
script processing.

Best regards,
Francesco Castellano