[OpenSER-Users] Parallel forking to loopback problem - for ringing multiple numbers at same time, simulring

Dave Singer dave.singer at wideideas.com
Tue Dec 4 22:12:52 CET 2007


by the way, I'm trying this using openser 1.3.0-pre1-notls from latest svn
of trunk as of 12-4-07

I'm trying to do findme logic..
Call comes in, lookup settings for where it is supposed to go.
>From the settings I find that the user wants it to ring a friends home
phone, a cell phone, and the home ata associated with account. To keep
voicemail handled by the home account and not by one of the places that they
are forwarding to they want voicemail to be triggered after 15 seconds.
The friends home phone is actually one of our accounts and has two ata's
associated with it that register as the same sip user and they have it also
ring a cell phone and the account sends the call to voicemail after 18
seconds.

Psudo Call flow/Logic of what I have so far.
Call comes in to openser for 555/123-4567
pull from usr_preferences settings for the call destination number.
so I get a stack of avps $avp(s:brahch_uri) from usr_preferences like this:
  sip:5551239876 at 127.0.0.1
  sip:localuser3 at 127.0.0.1
  sip:5553216548 at 127.0.0.1
I do diversion header then branch for them with:
  avp_pushto("$br", "$avp(s:branch_uri)"
I drop the local branch.
The branched calls come in as expected and verify that it is a valid
intentional loopback by testing:
  if ( ruri =~ "127.0.0.1" && dst_ip == 127.0.0.1 ) { < mark as valid
loopback > }
Like the beginning of the original call I lookup where they should go and if
the user in the diversion header is allowed to send them there (PSTN)
In this case
  one call would be sent out through the PSTN gateways with full failover
logic,
  another would call the device associated with the account by doing a
lookup("location")
  and the third, (the one to the friend that our customer) would add another
diversion header and start another parallel fork with loopback for the cell
phone and two ata's

When I test with a simple forward to a single PSTN number,
in the loopbacked session, I rewrite the ruri accordingly with real dest
user and real ip address.
the ruri shows what I want in the logs just before t_relay and exit and in
branch_route[1]
after last log entry from branch_route, I have in the log:
  ERROR:core:udp_send: sendto(sock,0xb619d954,1376,0,0xb619ba6c,16): Invalid
argument(22)
  CRITICAL:core:udp_send: invalid sendtoparameters one possible reason is
the server is bound to localhost and attempts to send to the net
  ERROR:tm:msg_send: udp_send failed
  ERROR:tm:t_forward_nonack: sending request failed
Then I have log entrys from onreply_route[1]:
  Reply - S=100 M=INVITE RURI=<null> src_ip = 127.0.0.1
  Reply - S=477 M=INVITE RURI=<null> src_ip = 127.0.0.1
not the real world IP address of the other server I set in the ruri
and there is nothing in the logs on the intended server or the local server
indicating routing logic for an invite failed

I'm woundering if the TM module, while processing the invite in the
loopback, is matching up the transaction with the session that originated
the loopback and using the settings and status from it.


Any Ideas, Thoughts, HELP!!?
Or is OpenSER just not ready for this yet?

Thanks Dave
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sip-router.org/pipermail/sr-users/attachments/20071204/90b6a1f6/attachment.htm>


More information about the sr-users mailing list