No time to test it, so give it a
try. If you are using 4.0 (even 3.3)
should
work by just cherry-picking. Let me
know if works fine.
Depending on feedback will be part
of 4.0.1 (out in few hours), or let
to be
backported for 4.0.2.
Cheers,
Daniel
On
4/25/13 2:23 PM, Uri Shacked
wrote:
Hi,
As
i wrote, when the BYE is
"normal" not
generated form
dlg_timeout. srv2 that
forwards it, does it OK.
The
TO/From headers in my
scenario are the same as
the
initial request in both
servers (there are 2 legs
for each call (incoming
and
outgoing), on each leg i
use the same initial
headers.Ii hope i was more
clear
now.
IMO
the timeout-triggered BYE
should have identical
From/To headers as the
initial request -
otherwise it may cause
issues with
clients that are strict,
or rely on identical
headers (like in your
setup).
Following this issue,
and the issue "BYE
dialog timeout bad
syntax"
from
Julia.
I have 2 kamailio
servers. srv1 creates
a dlg_timeout BYE and
sends it
to srv2 that forwards
the BYE to the next
sip server (some other
server...).
I can definitely see
that the BYE that is
being forward is
malformed.
The TO header info is
not complete (the port
section for example).
When a "normal" bye is
received from srv1
(not dlg_timeout) all
works
fine.
One thing that i do in
this scenario, on srv2
i use uac_replace() on
the
TO header. So, it is
OK that the final
destination is
different (i do
translations).
I attached the BYE
from both servers.
Any ideas?
Thanks,
Uri
You probably meant the
To and From headers,
I'm guessing. Yeah, as
long
as there is no display
name component, URIs
in headers like that
don't
have to go in
<>s. The way
Kamailio sends it is
grammatically valid,
per
RFC 3261. The user
agents are at fault
for not understanding
it
correctly. Alex
Balashov <abalashov
at evaristesys.com