[asterisk-users] SIP Problems continue...

[Olle E Johansson]

9 maj 2007 kl. 18.14 skrev Ken Williams:

> SIP channel hang ups are progressively getting worse and I'm really  
> grasping at straws here trying to find out what the cause is.  The  
> problem start, once a week or so the SIP phones couldn't  
> communicate with the server, though there was no error message on  
> the server and everything appeared fine on the server.  It's now  
> doing it multiple times a day and I fear having to go back to our  
> old phone system if I can't find a fix in the near future.  When  
> the SIP channel locks up the only fix is to restart Asterisk.  SIP  
> RELOAD & RELOAD CHAN_SIP do no good.
>
> Here's a few things I've noticed and changes I've made in hopes of  
> making it better.  First, I've currently got 71 active SIP channels  
> when only 2 people are on the phone.  This doesn't happen every  
> time, but could be part of the cause.  The 'ghost' channels are all  
> INVITES, how do I clear these without rebooting the system?
>
> 10.200.26.116    716         0a2a959d3d3  00102/00000  unkn   
> No       Init: INVITE
> 10.200.26.115    715         1dee947d485  00102/00000  unkn   
> No       Init: INVITE
> 10.200.26.104    704         28808764699  00102/00000  unkn   
> No       Init: INVITE
> 10.200.26.104    704         36d3e88f59c  00102/00000  unkn   
> No       Init: INVITE
> 10.200.26.104    704         0e00060800d  00102/00000  unkn   
> No       Init: INVITE

There is an open bug report on this in the bug tracker already.

I need your help to find what's causing this issue and provided I can  
get proper information from you,
will spend time locating the bug.

First, enable SIP history and catch history for these calls that hang  
with "sip show history"

Secondly, check the dialplan and tell me more. Where are you calling,  
why doesn't the other end respond?
It's usually calls where we retransmit a number of times and then  
forget to destroy the calls.

If I can get a better description so I can repeat this, I'm sure the  
bug can be killed.

In the bug tracker, there's a patch that will help you. However,  
until I find more exact information
about the nature of these calls, I'm unwilling to commit it. To  
commit a fix to a poorly defined issue
is usually causing more issues, something I can do in trunk but don't  
want to do in release code.

Please send the required information directly to my e-mail address  
and I'll take a look.
Thank you for your assistance with this bug.

/Olle