asterisk users - Sep 2005 - Bad FCS nightmare to Nortel SL100 with TE410P

I have an * box connected to a Nortel SL100 through a PRI (US) using the
Digium TE410P (quad-span T1 card).  I don't have access to the SL100 -
it is handled by another group.

The span comes up OK (timing, framing fine).  However, as soon as the
D channel comes up, I get endless "HDLC Bad FCS" errors.  I modified
logger.conf to get rid of the messages (so I could see what else was
going on), and noticed that the B-channel restart was going horribly
slow, and the D channel was essentially "flapping" up and down.  I
could sometimes squeeze a call in while the D channel was up, but
it would only last a few seconds.  I also get "short write" errors
as well (unfortunately I don't have a log of these and can't get at
the PRI at the moment to get the exact error message).

I've had the physical circuit tested and there are no issues with it.
In fact, it was working fine to the same switch as an E&M digital trunk
up until we tried to change it to a PRI.

I've tried 3 different TE410Ps on three different * versions (based
on things I've seen in previous posts).  All behave exactly the same.
The versions are 1.0.5, 1.0.9, and a CVS version of 1.2.0-beta1 pulled
down at the end of August.

In all cases, the systems are Dell PowerEdge 1750s (using RAID, no
IDE drives involved) on Debian / kernel 2.4.27.  I see no indication
of problematic interrupts.  In one test, there were 3 other PRIs running
on the TE410P (in production) and there are no problems with any other
PRIs.  Ditto the configuration (I've checked and am doing the exact
same thing with all my PRIs, just on different channels).

Before I start providing configuration excerpts - has anyone had this
problem connecting to an older Nortel Meridian switch and if so, what
did you do to fix it?  I suspect that there is a subtle configuration
option on the SL100 that is wrong, but since I don't have access to
it I can't confirm that.  Can the wrong switch type cause FCS errors?
Is there anything specific I can look at?  For those who speak SL100,
do you know of any specific parameter I can point the SL100 guy to?

One more data point:  I threw the PRI from the SL100 onto a spare
port on a Cisco AS5350 and the AS5350 isn't complaining (no frame slips,
no problem with the D channel).

I'm pulling my hair out with this.  Any help or pointers to info would
be helpful.  I will post a summary to the list if I get any useful
private e-mail about this.

Thanks!

---
Gil Kloepfer
astr-usr@kloepfer.org

On Tue, Sep 27, 2005 at 01:22:48AM -0500, I wrote:
> I have an * box connected to a Nortel SL100 through a PRI (US) using the
> Digium TE410P (quad-span T1 card).  I don't have access to the SL100 -
> it is handled by another group.
> 
> The span comes up OK (timing, framing fine).  However, as soon as the
> D channel comes up, I get endless "HDLC Bad FCS" errors.
We have resolved the issue - it turned out that there was something
wrong at the SL100 side of the connection.  The bad news is that I
can't get anyone who manages that SL100 to tell me exactly what
the problem actually was.  The best I was able to get was, "It was
some kind of cable or configuration problem."

The only private feedback I got were recommendations to try the
Sangoma cards (it looks like they keep more in the way of statistics).

I also have no idea why the TE410P complained, but the Cisco AS5350
didn't.

Thanks to everyone who provided feedback.

---
Gil Kloepfer
astr-usr@kloepfer.org