[Serdev] [BUG] Closed: (SER-26) Investigate the problem with outbound unicast sockets and multicast destination IP

[Jan Janak (JIRA)]

     [ http://bugs.sip-router.org/browse/SER-26?page=history ]
     
Jan Janak closed SER-26:
------------------------


> Investigate the problem with outbound unicast sockets and multicast destination IP
> ----------------------------------------------------------------------------------
>
>          Key: SER-26
>          URL: http://bugs.sip-router.org/browse/SER-26
>      Project: SIP Express Router
>         Type: Bug
>   Components: core
>     Versions: HEAD
>     Reporter: Jan Janak
>     Assignee: Jan Janak
>      Fix For: HEAD

>
> Hi,
> I am picking up where my collegue Samuel left it. I don't send it to
> the list, as i am including some long files of logging and parts of the
> config file.
> I have tried checking the effective values of mcast_loopback after
> setsockopt with getsockopt, and it is correctly set.
> I also disabled IP_MULTICAST in the kernel source files, so by default,
> loopback is disabled.
> Also, I changed the order of listen statements in the config file.
> No luck with any of them.
> We've been discussing it among us, and we have a possible reason for
> this. SER opens two sockets (one on the unicast, one on the multicast),
> each a different process (right?). The multicast socket is set to have
> the loopback disabled. But, when the SIP message is sent (forward,
> t_relay_to_udp) to multicast, the UNICAST SOCKET IS USED. So, the kernel
> is not aware of these two sockets belonging to the same
> interface/machine, so it forwards the SIP message to the multicast
> socket, which doesn't filter it out as being loopbacked because the
> multicast socket is not the actual sender, thus, it is not a loopbacked
> message.
> As an attachment, i include a log file, where a debug=4 is used (just
> deleted some chunks that are too big ... ). The two invite messages can
> be seen there (first message starts at line 650, the second at line
> 750). Note that the second has an extra header that we add for debugging
> purposes (P-hint: routed to multicast). The source ip address is printed
> on both: one comes from 192.0.1.49 (the phone), the second comes from
> 192.0.1.250 (SER, the sender into multicast).
> I also attach the relevant parts of the config file. See that i tried
> with forward and t_relay_to_udp ...
> The detection of the loopback message right now we do it in the config
> file ... and though not the desired solution, it works. But maybe, there
> is a way to solve it in the SER code itself. If the problem is as
> described above, maybe there could be a way to choose via which socket
> the message is to be sent ... so, if the message is going to be
> multicasted, use the multicast socket already opened, thus the loopback
> filter can do its job. Just a thought.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
   http://bugs.sip-router.org/secure/Administrators.jspa
-
If you want more information on JIRA, or have a bug to report see:
   http://www.atlassian.com/software/jira