[asterisk-dev] meetme and DTMF going from Zap to SIP-IAX2

[Matt Florell]

On 8/10/07, Donny Kavanagh <donnyk at gmail.com> wrote:
> 1.2.x is totally unsupported now except for security fixes, we would
> much rather you helped us reproduce and fix those bugs rather then
> back porting features you need.

I know it is unsupported, but it is much more stable, and incurs less
load on the server doing the same things. I do understand that the
bugs need to be fixed, and I will help you reproduce if you have the
time.

Here are load graphs of the two days on the same system, yesterday
with 1.4.10 and today with 1.2.23.
http://www.eflo.net/images/Asterisk-1.4.10_load.gif
http://www.eflo.net/images/Asterisk-1.2.23_load.gif
* please note the graphs are in a different scale, but the number of
calls handled and the talk time are very close for both days.

There was a crash after 10AM on 1.4.10, and no crashes today on 1.2.23

You can see the load spikes on 1.4.10, and the spikes of channel
counts where the "core show channels concise" generated a lot of extra
output on non-existent duplicate channels.

> If you can reproduce the problem on a test system and can give access,
> russell would be more then happy to help you.

I can reproduce the problem on any system that has VICIDIAL installed
and can be put into performance testing mode.

Thanks,

MATT---


> Donny
>
> On 8/10/07, Matt Florell <astmattf at gmail.com> wrote:
> > Hello,
> >
> > Just wanted to follow up on this string, After several issues with
> > 1.4.9(IAX out of threads at 8 IAX channels) I upgraded to 1.4.10 as
> > soon as it was released which was better, but still had issues with
> > consistency of  the "show channels concise" output(channels with wrong
> > data, duplicates-triplicates-quadruplicates-then no results at all)
> > and the load of 1.4.10 was almost double that of 1.2.23 at the same
> > call load on the same hardware when going above 70 channels.
> >
> > In the end I downgraded back to 1.2.23 and did a hack of routing the
> > SIP channel to a stand-alone machine with a quad T1 card and two T1
> > ports crossed-over to each other. It took a couple hours to throw that
> > together, and it is an ugly solution, but it works.
> >
> > The calls go out of the meetme on server A, over SIP trunk to server B
> > where they loop out of one T1 to the other and then to an AGI script
> > where the original server-A T1-connected user's tones are detected.
> >
> > I plan on working on trying to backport the SLA features within meetme
> > to the 1.2.X tree and will post on the tracker if I am successful.
> >
> > MATT---
> >
> >
> > On 8/6/07, Russell Bryant <russell at digium.com> wrote:
> > > Matt Florell wrote:
> > > > I looked at the code and it is pretty drastic how much has changed in
> > > > app_meetme.c since the 1.2.X tree. I tried the "F" flag with 1.4.9 and
> > > > it works great with every combination of channels I could throw at it
> > > > passing DTMF through. I am now going to start testing 1.4.9 under some
> > > > load to see if I can use it in production.
> > >
> > > That's great news that it works well for you.  Thanks for letting me know.
> > >
> > > --
> > > Russell Bryant
> > > Software Engineer
> > > Digium, Inc.
> > >
> > > _______________________________________________
> > > --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
> > >
> >
> > _______________________________________________
> > --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
> >
>
> _______________________________________________
> --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
>