Digium b410p with i4b

[Hans Petter Selasky]

Hi Mikael,

On Thursday 12 July 2007 23:50, Mikael Batard wrote:
> On 12.07.2007 23:19, Hans Petter Selasky wrote:
> > Hi Mikael!
>
> Hey Hans,
>
> thanks for your fast answer :)
>
> > Se comments below.
> >
> > I cannot see any ALERT message in the log only a CONNECT.
> >
> > The Ringing command should be first, before Answer !
>
> It is

Ok.

>
> > And it should not connect the B-channel.
>
> The ringing command ?

Yes.

>
> > This connect is too early. We should have seen an ALERT message.
>
> So, this is the bug which causes the problem ?

Not according to the timing in the log.

>
> > Now, this is strange: This is the response to the second STATUS enquiry.
> > But why is is processed so late?
>
> Maybe something with the pcm ?
> I'm using the configuration of isdnconfig from the manpage, but I didn't
> know if it's the right thing to do.

No, that is not the problem.

>
> > The PBX had enough! I probably wants to terminate the dummy
> > callreference, but instead it terminates another call! This can maybe be
> > exploited.
>
> Maybe because I've made another call just before this one, to reach a
> missed one with the debug ?

If you look at the Callreference value, you see that the PBX tries to 
terminate the last incoming call multiple times. This looks like a bug in the 
PBX!

>
> > I4B follows up with a RELEASE_COMPLETE, though it is strictly not
> > required, but does send it anyway. This is not a violation of the ISDN
> > protocol, hence during a race when both ends wants to terminte at the
> > same time, this can happen.
>
> OK
>
> > This is suspicious: Why does the PBX send a double RELEASE_COMPLETE. I
> > see no reason for this.
>
> Maybe the other call before this one ?

I think the PBX (or your Teleco provider) is about to hang. You can expect 
that it will run out of memory and stop working after X such calls :-)

>
> > This is also strange. Now the PBX wants to terminate the call again, but
> > as you see, the TEI value is completely wrong. TEI 0 is the TEI value for
> > Point2Point. Also the location value is wrong.
>
> Yes, duno :(

It is a PBX bug.

>
> > Can you tell us a little more about your setup. Is the HFC card connected
> > to an ISDN PBX or directly to the ISDN telecom provider?
>
> Of course.

Directly to the Telecom provider. Ok.

>
> So, i'm using asterisk 1.2.17, asterisk-addons 1.2.3, zaptel 1.4.1 (for
> tdm400p card, with 2 fxo & 2 fxs) : all from ports
> I'm using i4b and chan_capi from svn trunk, revision 516.
> The HFC card is a b410p (quadbri) from digium and is connected to our
> ISDN telecom provider.
> There is another ISDN PBX (old alcatel one) connected (to our ISDN
> telecom provider) in parallel to the asterisk one, which works fine (but
> it's old and we're using a sip trunk with a voip provider, so we need to
> use an asterisk one). All calls are handle normally by this old PBX.
> We have seen that there were missed calls, because only the old PBX has
> made ringing the phones when it happens.
>
> > It looks to me like your PBX or Telco has a weak point in their code. I'm
> > not sure if I want to make another workaround in I4B for this kind of
> > things, hence I already have several workarounds for various PBX bugs.
> > Let's hope it is just your PBX :-)
>
> Hope too :)

--HPS