[nsp] AS53xx V.92 configuration not working

Cisco Geek Rotation cisco@peakpeak.com
Mon, 19 Aug 2002 15:07:21 -0600


--=====================_61035434==_.ALT
Content-Type: text/plain; charset="us-ascii"; format=flowed

At 12:45 PM 8/19/2002 -0700, Aaron Leonard wrote:
>The modem autoconfigure code is ALMOST platform independent (except
>that the 5800 [and maybe the 5850?] does some task queueing for
>scalability's sake.)  We have seen some autoconfig failures on
>the high-end platforms (5800, 5850) for very long config strings,
>which we believe we can work around by boosting autoconfigure's
>read timeout (CSCdy00247).
>
>However none of this would seem to be relevant to the issue raised
>below, since it's talking about a 5300, not a 58xx, and about a
>short modemcap (e.g. &F&D2S63=3), not a long one.  So I believe
>that this is not a modemcap issue but a modem AT syntax issue -
>so let's zoom into the particulars.
>
>Chris has an AS5300 running 12.2(2)XA5 and MICA portware 2.9.2.0.
>He's trying to send a modemcap of:
>&F&D2S62=8S63=3S21=15S29=12S34=18000S40=10S54=456S10=50S71=4, but this is
>(I presume) being rejected by the MICA AT parser.
>
>I think I'll spare everyone an exegesis of these S registers
>(I suspect that the cisco-nsp folk in particular would find the
>topic sub-fascinating).
>
>Here are the modemcaps I'm currently recommending for MICA 2.9.2.0:
>
>   Recommended modemcap for 2.9.2.0 in a V.92/V.44 application:
>   &F&D2S54=16584S0=0S29=12S21=15S62=8S63=3S34=18000S40=10S10=50

Thanks.

That string works (at least, the modem initialization error messages go 
away). But the calltracker database verbose log messages still say V.90 in 
them.  I imagine this is a bug too?  How do you tell if V.92 is established?

The modem is a USRobotics V.92 modem (upgraded from V.90). ATI7 lists V.92 
as one of its options. It connects at 46-52K, loop length is 691 feet from 
the Central Office.

I tried debug modem csm though and here is what I got:

Aug 19 20:43:58.952: Mica Modem(1/25): State Transition to Quick Connect

So I assume that V.92 is working, I just don't know how to make the box 
tell me it is. The connect times to authentication are on the order of 20 
seconds, so this seems faster than a "standard" V.90 connect.

I'm seeing plenty 'o weird things on this box, like a modem that isn't 
listed as B or T status, but continually tries to get initialized over and 
over and over.

Keep working on that code! :)

Chris
--=====================_61035434==_.ALT
Content-Type: text/html; charset="us-ascii"

<html>
At 12:45 PM 8/19/2002 -0700, Aaron Leonard wrote:<br>
<blockquote type=cite class=cite cite>The modem autoconfigure code is
ALMOST platform independent (except<br>
that the 5800 [and maybe the 5850?] does some task queueing for<br>
scalability's sake.)&nbsp; We have seen some autoconfig failures on<br>
the high-end platforms (5800, 5850) for very long config strings,<br>
which we believe we can work around by boosting autoconfigure's<br>
read timeout (CSCdy00247).<br><br>
However none of this would seem to be relevant to the issue raised<br>
below, since it's talking about a 5300, not a 58xx, and about a<br>
short modemcap (e.g. &amp;F&amp;D2S63=3), not a long one.&nbsp; So I
believe<br>
that this is not a modemcap issue but a modem AT syntax issue - <br>
so let's zoom into the particulars.<br><br>
Chris has an AS5300 running 12.2(2)XA5 and MICA portware 2.9.2.0.<br>
He's trying to send a modemcap of:<br>
&amp;F&amp;D2S62=8S63=3S21=15S29=12S34=18000S40=10S54=456S10=50S71=4, but
this is<br>
(I presume) being rejected by the MICA AT parser.<br><br>
I think I'll spare everyone an exegesis of these S registers<br>
(I suspect that the cisco-nsp folk in particular would find the<br>
topic sub-fascinating).<br><br>
Here are the modemcaps I'm currently recommending for MICA
2.9.2.0:<br><br>
&nbsp; Recommended modemcap for 2.9.2.0 in a V.92/V.44 application: 
<br>
&nbsp;
&amp;F&amp;D2S54=16584S0=0S29=12S21=15S62=8S63=3S34=18000S40=10S10=50
</blockquote><br>
Thanks.<br><br>
That string works (at least, the modem initialization error messages go
away). But the calltracker database verbose log messages still say V.90
in them.&nbsp; I imagine this is a bug too?&nbsp; How do you tell if V.92
is established?<br><br>
The modem is a USRobotics V.92 modem (upgraded from V.90). ATI7 lists
V.92 as one of its options. It connects at 46-52K, loop length is 691
feet from the Central Office.<br><br>
I tried debug modem csm though and here is what I got:<br><br>
<font face="Courier New, Courier">Aug 19 20:43:58.952: Mica Modem(1/25):
State Transition to Quick Connect<br><br>
</font>So I assume that V.92 is working, I just don't know how to make
the box tell me it is. The connect times to authentication are on the
order of 20 seconds, so this seems faster than a &quot;standard&quot;
V.90 connect.<br><br>
I'm seeing plenty 'o weird things on this box, like a modem that isn't
listed as B or T status, but continually tries to get initialized over
and over and over.<br><br>
Keep working on that code! :)<br><br>
Chris</html>

--=====================_61035434==_.ALT--