I am having a problem with calls that are parallel forked. The problem happens when a call comes in, gets forked using multiple aliases and one UA is unavailable. The question is, how do I handle the offline call leg?
I can't do if (!lookup("location")) { break; (or sl_send_reply("404") } because then the whole call dies since each branch of the call uses the same callid.
If I t_relay them both, then the fork works but I the offline leg gets repeatedly relayed unltil I get a 483 "Too many hops". This scenario would be OK [althought certainly not ideal], except it does not allow me to add a condition to hunt to the next number upon a failed lookup("location"). If I do, the online UA will ring once and the offline UA will get routed immediately to voicemail which picks up after just one ring.
How are other handling these problems? I think the answer lies in letting failure_route handle everything, but I don't know how to get that to work because if I don't handle the offline users in the (!lookup("location"), I end up with routing loops and eventualluy a 483. Anybody that has no online contacts whatsoever never makes it to failure_route (where the forward to VM is handled).
Below is my ser.cfg. I've been playing around with things like if (isflagset(6)) { route(x) } but I still need some way to handle the failed portion of the fork. Is there a way to "null route" the failed portion or something?
route {
/* Registration and other logic snipped for brevity */
# Translate local address according to aliases table xlog("L_INFO", "%ci: Preparing to look up alias for %ru\n"); if(lookup("aliases")) { xlog("L_NOTICE", "%ci: alias lookup changed uri to %ru\n");
if (!(uri==myself)) { xlog("L_NOTICE", "%ci: Outbound aliases %ru\n"); route(4); break; };
};
if ( (uri=~"^sip:411@.*") | (uri=~"^sip:911@.*") | (uri=~"^sip:011[0-9]+@.*") | (uri=~"^sip:1[0-9]{10}@.*") ) { route(3); # 3: PSTN with authentication break; };
# Anything left is a local call.
xlog("L_INFO", "%ci: Preparing to look up location for %ru\n"); if (!lookup("location")) { xlog("L_NOTICE", "%ci: no location for %ru\n"); setflag(5); } else { # Had a sucessful lookup setflag(6); };
route(4);
}
route[4] { # If an invitation, we want to hunt on failure
if(method == "INVITE") { t_on_failure("1"); # first hunt };
append_hf("P-hint: relay\r\n");
xlog("L_NOTICE", "%ci: r4: t_relay %ru\n"); if (!t_relay()) { xlog("L_NOTICE", "%ci: %ru r4: Relay Failed\n"); sl_reply_error(); break; };
}
failure_route[1] { revert_uri(); route(7); } ### Handle hunt sequences.
I am not necessarily looking for somebody to tell me how to fix my code, more for somebody to suggest how it might be done or how they have done it.
Thoughts:
1. Is there a way for one branch to check the status of another?
2. Do you think it is possible with AVP to lookup from original ruri and then append_brach only if lookup suceeds?
On 8/3/05, Daniel Poulsen dpoulsen@gmail.com wrote:
I am having a problem with calls that are parallel forked. The problem happens when a call comes in, gets forked using multiple aliases and one UA is unavailable. The question is, how do I handle the offline call leg?
I can't do if (!lookup("location")) { break; (or sl_send_reply("404") } because then the whole call dies since each branch of the call uses the same callid.
If I t_relay them both, then the fork works but I the offline leg gets repeatedly relayed unltil I get a 483 "Too many hops". This scenario would be OK [althought certainly not ideal], except it does not allow me to add a condition to hunt to the next number upon a failed lookup("location"). If I do, the online UA will ring once and the offline UA will get routed immediately to voicemail which picks up after just one ring.
How are other handling these problems? I think the answer lies in letting failure_route handle everything, but I don't know how to get that to work because if I don't handle the offline users in the (!lookup("location"), I end up with routing loops and eventualluy a 483. Anybody that has no online contacts whatsoever never makes it to failure_route (where the forward to VM is handled).
Below is my ser.cfg. I've been playing around with things like if (isflagset(6)) { route(x) } but I still need some way to handle the failed portion of the fork. Is there a way to "null route" the failed portion or something?
route {
/* Registration and other logic snipped for brevity */
# Translate local address according to aliases table xlog("L_INFO", "%ci: Preparing to look up alias for %ru\n"); if(lookup("aliases")) { xlog("L_NOTICE", "%ci: alias lookup changed uri to %ru\n");
if (!(uri==myself)) { xlog("L_NOTICE", "%ci: Outbound aliases %ru\n"); route(4); break; };
};
if ( (uri=~"^sip:411@.*") | (uri=~"^sip:911@.*") | (uri=~"^sip:011[0-9]+@.*") | (uri=~"^sip:1[0-9]{10}@.*") ) { route(3); # 3: PSTN with authentication break; };
# Anything left is a local call.
xlog("L_INFO", "%ci: Preparing to look up location for %ru\n"); if (!lookup("location")) { xlog("L_NOTICE", "%ci: no location for %ru\n"); setflag(5); } else { # Had a sucessful lookup setflag(6); };
route(4);
}
route[4] { # If an invitation, we want to hunt on failure
if(method == "INVITE") { t_on_failure("1"); # first hunt };
append_hf("P-hint: relay\r\n");
xlog("L_NOTICE", "%ci: r4: t_relay %ru\n"); if (!t_relay()) { xlog("L_NOTICE", "%ci: %ru r4: Relay Failed\n"); sl_reply_error(); break; };
}
failure_route[1] { revert_uri(); route(7); } ### Handle hunt sequences.
-
Daniel Poulsen