[Asterisk-Dev] chan_sip cvs-stable post 1/1/2005 polycom bug

Rich Adamson radamson at routers.com
Sat Jan 15 00:10:57 MST 2005


> Hello,
> 	I've opened a bug that is affecting Polycom phones. As far as I 
> can tell, the problem was introduced in chan_sip between 1/1/2005 and 
> 1/2/2005, and appears in subsequent versions.
> 
> http://bugs.digium.com/bug_view_page.php?bug_id=0003348
> 
> Polycom Soundpoint IP 300 and 500 phones running both the 1.3.1 and 1.3.4 
> SIP firmware are unable to complete calls with any cvs-stable chan_sip 
> after 1/1/2005. Calls appear to complete, and the first frame of the audio 
> is heard, but the Polycom then gives a fasy busy.
> 
> I have spent most of this evening tracking this down. I first noticed the 
> problem with an update to a client's server on 1/8/2005. All their Polycom 
> phones exhibited the same behaviour and were unable to make any outbound 
> calls. I then rolled chan_sip back to the 12/26/2004 version and it worked 
> fine.
> 
> I called Digium and spoke with Matt and asked him if he had any reports of 
> this, and he said no. I promised that I would try to track it down, update 
> my firmware, check all configuration settings and open a bug report if 
> neccessary.
> 
> Tonight, I spent a good portion of time actually testing various chan_sip 
> code and found that the problems are not present in the chan_sip code 
> before 1/2/2005. I am able to reproduce this faithfully, and have 
> attempted to discuss this on #asterisk, but have not been able to catch 
> Russell or Mark.
> 
> I have attached a diff of the changes between 1/1/2005 and 1/2/2005, and 
> the changes are minor, mostly related to mailbox checks. I haven't dug 
> into the code to find more detailed information, but I will attach 
> tcpdumps from both a successful 1/1/2005 call and an unsuccessful 1/2/2005 
> call. Hopefully, we can figure out where this issue is before 1.0.4 is 
> dropped with this problem, and avoid a lot of complaints on the list about 
> Polycom phones not working with it.

Could that be the fix applied with bug #3129 where progressinband=no
was added?





More information about the asterisk-dev mailing list