[Asterisk-Users] chan_h323 + openh323 CVS = no go?

[Siggi Langauf]

On Tue, 10 Jun 2003, Jeremy McNamara wrote:

> >trying to build the h323 channel driver that comes with asterisk works
> >fine, but only as long as I use openh323-1.11.7.
> >
> >Unfortunately, that setup seems to have a bug which misguides one of the
> >audio streams. (So while * can "hear" me, the phone remains silent.)
>
> Open H.323 1.11.7 works perfectly in all of my installations. I babysit
> 15 different chan_h323 based systems.
>
> One way audio usually means you have codec problems or are trying to
> traverse NAT.

Nope, things were more weird in this case: This installation has a few
hundred Cisco 79xx phones running in Skinny mode babysitted by a Cisco
CallManager (actually two CCMs, if you count the fallback machine).
Asterisk is used as a voicemail box attached to the CCM as an H.323
gateway. So what happens is: the CCM builds all connections to asterisk
but negotiates via H.245 that the actual voice streams should be sent
directly to the phone. For some reason, OpenH323 1.11.7 would ignore this
and just send packets to the CCM instead, which would just drop them.
Hence silence on the phone. The "codec" is G.711, so no problems here. and
everything's running in one big private class B net without any outside
connection, so no NAT.

[...]
> It looks like the Open H.323 folks either forgot to include the G.729
> Capability stubb or were forced to pull it by their legal department.
> I will look into this.

It's still there, but skipped during compilation. Why, I can't tell.

> >Is that driver not supposed to work with current OpenH323??
> >Anything I'm doing wrong?
> >
> We have never tested the latest cvs -HEAD of Open H.323 and PWLib, as
> there have been major changes, so we are giving those guys some time to
> make sure everything is stable before we dive in to new, untested code.

I see.
So I'll have to stick to chan_oh323 for now.

Thanks,
	Siggi