[Asterisk-Dev] Not for 1.2, was SIP RTP JitterBuffer in Asterisk - now aimed for 1.3 dev

Roy Sigurd Karlsbakk roy at karlsbakk.net
Wed Sep 14 02:27:16 MST 2005


hi

i did some live testing with head and sip rtp jb, and got a core dump  
when the load climed.

would it be possible to sponsor this project to speed up the  
development?

roy

On 26. aug. 2005, at 15.48, Zoa wrote:

>
> We did a lot of testing on the jitterbuffers, an i can only confirm  
> that
> (both the adaptive version and the fixed version) use up a lot of cpu
> resources, (memory is not so bad).
> When you go to 99% cpu utilization, there is some memory leak (as long
> as you stay under 95% there is no memory leak). We have no clue if  
> this
> is in our code or in the main asterisk code, we do know its not in the
> jitter buffer code itself as we have it with both the stevek  
> version as
> the slav version. (The slav version was created to make sure the  
> problem
> was not in the stevek version).
>
> So you definately need to be able to turn it off (you can already, but
> not per user/peer).
>
> Zoa.
> -----
>
> http://www.asteriskguru.com
>
>
>
> Steve Kann wrote:
>
>
>> Matt wrote:
>>
>>
>>> I'm going to assume it's still a patch based on what I've seen  
>>> elsewhere.
>>> As far as responding to arnaud I have to say "uhh why" as well.  Why
>>> would you not want the jitter buffer on all the time for all users?
>>> It really can only help (if it works correctly) which so far it does
>>> indeed seem to.
>>>
>>>
>>>
>> The present patch implementes a fixed length buffer of some sort.   
>> So,
>> even if it's working right, it will add some delay and resource usage
>> (CPU, memory).  So, some people might not want to use it.
>>
>> The original adaptive buffer could theoretically be a lot better  
>> about
>> not adding excessive delay when no jitter is present, but it still
>> adds CPU and some memory overhead.
>>
>> Both implementations add complexity to the media path, and can
>> theoretically cause issues, especially if the sender is sending
>> incorrect timestamps, or has other bugs.  Ideally, we'd be able to
>> diagnose and work around these bugs in the remote clients, but that
>> takes time and effort, and the quickest short-term solution might be
>> to just turn it off.
>>
>> So, I could see why people would want the ability to turn it off.
>>
>> -SteveK
>>
>>
>>> On 8/26/05, Matt <mhoppes at gmail.com> wrote:
>>>
>>>
>>>
>>>> I'm confused now.. is the sip jitter buffer currently available  
>>>> in the
>>>> CVS-HEAD or does it still need to be patched in?
>>>>
>>>> On 8/26/05, Arnaud <arno at directcentrex.com> wrote:
>>>>
>>>>
>>>>
>>>>> Eric Wieling aka ManxPower wrote:
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>> Kevin P. Fleming wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>> AEL (pbx_ael.c) will be included in the 1.2 release, but will be
>>>>>>> clearly marked in the UPGRADE.txt file as experimental. If we  
>>>>>>> don't
>>>>>>> include it in 1.2, we won't get very many testers other than the
>>>>>>> brave souls who will continue to run the development branch :-)
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>> The same could be said about the SIP jitter buffer.
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>> I agree, and also it will be better to be able to enable SIP  
>>>>> jitter per user
>>>>> --
>>>>>
>>>>> Arnaud Pignard (apignard at frontier.fr)
>>>>> Standard : + 33 1 70 71 50 00 - Fax : +33 1 70 71 50 60
>>>>> MSN : stormfr at hotrmail.com - ICQ : 20946060
>>>>>
>>>>> Frontier Online - Opérateur Internet - http://www.frontier.fr
>>>>> Direct Centrex - Opérateur Télécom sur IP - http:// 
>>>>> www.directcentrex.com
>>>>> Direct Nom - Registrar de nom de domaine - http:// 
>>>>> www.directcnom.com
>>>>>
>>>>> _______________________________________________
>>>>> Asterisk-Dev mailing list
>>>>> Asterisk-Dev at lists.digium.com
>>>>> http://lists.digium.com/mailman/listinfo/asterisk-dev
>>>>> To UNSUBSCRIBE or update options visit:
>>>>>   http://lists.digium.com/mailman/listinfo/asterisk-dev
>>>>>
>>>>>
>>>>>
>>>>>
>>> _______________________________________________
>>> Asterisk-Dev mailing list
>>> Asterisk-Dev at lists.digium.com
>>> http://lists.digium.com/mailman/listinfo/asterisk-dev
>>> To UNSUBSCRIBE or update options visit:
>>>   http://lists.digium.com/mailman/listinfo/asterisk-dev
>>>
>>>
>>>
>>>
>>
>> --------------------------------------------------------------------- 
>> ---
>>
>> _______________________________________________
>> Asterisk-Dev mailing list
>> Asterisk-Dev at lists.digium.com
>> http://lists.digium.com/mailman/listinfo/asterisk-dev
>> To UNSUBSCRIBE or update options visit:
>>   http://lists.digium.com/mailman/listinfo/asterisk-dev
>>
>>
>
> _______________________________________________
> Asterisk-Dev mailing list
> Asterisk-Dev at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-dev
> To UNSUBSCRIBE or update options visit:
>    http://lists.digium.com/mailman/listinfo/asterisk-dev




More information about the asterisk-dev mailing list