[asterisk-dev] After a BYE, chan_sip need 32s to destroy a SIP dialog. bug or feature?
Leo Leo
lafacce at yahoo.com.br
Fri Sep 16 09:36:11 CDT 2011
Hi,
The difference is: Version 1.4 don't follow the SIP recommendation and version 1.8 does.
By the documment, after receiving a BYE, the dialog shall be destroyed after 32*T1 timer expires.
As I remmember, the recommendation doesn't say anything about SDP and UDP ports. In asterisk, this is only freed when the dialog ends.
Leonardo
________________________________
De: Kristijan Vrban <vrban.lkml at googlemail.com>
Para: Asterisk Developers Mailing List <asterisk-dev at lists.digium.com>
Enviadas: Sexta-feira, 16 de Setembro de 2011 11:21
Assunto: [asterisk-dev] After a BYE, chan_sip need 32s to destroy a SIP dialog. bug or feature?
Hello, i am doing stress testing 1.4 vs. 1.8 and the result is, that
in a varying
tests i have done, that (very very short summary) 1.8 need less CPU
then 1.4, also
1.8 does not have the 200 calls limit, were you get "ERROR[11264]:
utils.c:968 ast_carefulwrite: write() returned error: Broken pipe"
great! but when i was doing the stress tests, i run out of free udptl
ports because:
chan_sip in Asterisk 1.8 has the behavior, that i does not destroy the
SIP dialog immediately
after it get a BYE. So the UDP Ports for RTP and T.38 stay open, for
32s until the "Really destroying SIP dialog
'006E36FB-DADE-E011-BB63-0024E820D141 at 127.0.0.1' Method: BYE" message
arrive. So after every SIP call two UDP ports are blocked for 32sec.
chan_sip in 1.4 immediately destroy the SIP dialog after a BYE, and
freed the used ports.
Is this a bug or a feature?
Kristijan
--
_____________________________________________________________________
-- Bandwidth and Colocation Provided by http://www.api-digital.com --
asterisk-dev mailing list
To UNSUBSCRIBE or update options visit:
http://lists.digium.com/mailman/listinfo/asterisk-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-dev/attachments/20110916/126170af/attachment.htm>
More information about the asterisk-dev
mailing list