[Asterisk-bsd] Threading troubles 1.4.5 & IAX2-> SIP

[Richard E Neese]

currently 1.4.5 on bsd has issues.

1 because zaptel is out of sync asterisk does not see and include zaptel in 
its build. 1.4.5 seems to have the iax issue. Asterisk 1.4.6 should be 
released soon to fix the current issues.

I have svn of branches/1.4 and the sip and iax issues are not there but the 
zaptel issue still is. 

I have worked with killfill from the asterisk-bsd group to get the ports 
updated but we are waitng to hear back from Gonzo about zaptel..

asterisk addons also fails to build currently. we are working to fix this and 
get the ports updated give us a fe more days.

On Monday 25 June 2007 09:28:05 Hendrik Visage wrote:
> Hi there,
>
>  FreeBSD 6.2
>  Asterisk 1.4.5 (and 1.4.3 from ports)
>
> Sip phone - PBX(*) -IAX2-VROUTER(*)- SIP-Voip provider
> (SPA901 & SPA922 phones)
>
> We've see a situation where the IAX2 appears to "loose"/drop the voice
> data to be sent to the
> SIP side of things. This happens "semi" intermittently, but we can
> reliably regenerate it
> at >40 alaw calls (even on a dedicated 1G network) and also with G729
> (but a tad more calls).
> It appears to happen using both trunking and non-trunking modes.
>
>  This happened with DONT_OPTIMIZE setting on or off, but with it ON it
> doesn;t dump core.
> At least when it was dumping core, it appeared to have been in the
> pthread_cancel
> calls.
>
> We've recompiled the PBX asterisk with no threading, and the
> milliwat/etc. tests to the vrouter
> from the SIP phones ran clean (other than when we pushed the bandwidth
> limits <grin>)
>
>  This morning it was consistently the agent (on the SIP Phones) who
> could hear the remote side complaining that the remote side can't hear
> them anymore. After we've recompiled the VROUTER Asterisk with
> non-threading, the calls stayed stable.
>
>  What appears to happen is that somewhere in the threading the IAX
> voice data is discarded or something on the way to the SIP side.
>
>  Anybody else anything like this?
> Any other work around for this issue/problem??