Re: WAN Port wigs-out

Charles Scott (cscott@freeway.net)
Thu, 12 Oct 1995 12:47:44 -0400 (EDT)

Dennis:
Well, it looks like there's more comming out of the woodwork with this
problem. We've seen it here and are currently working with Livingston to
figure it out. The guy at the other end of the line from the one that
first gave us trouble had seen it before also, which is how we knew not
to bother calling the phone company again.
I thing you'll find that it works fine if you disconnect the buss cable
from the main board which goes to the other 2 10 port boards. You should
also be able to make it go crazy by pingin it with pattern "ff". What
made me do that was that it was having more problems with binary file
transfers and image files. Looks like lots of bits on give it trouble.
We are waiting for our rep at one site to reconnect the other 20 ports
so we can retest the unit with 3.1.4 (which Brian uploaded to the unit
the other day). We'll see if that does anything.
My personal guess is that it's a hardware problem, either ringing on
the data bus due to termination problems or problems with the drivers at
the WAN port or supply impeadance feeding those drivers. When we pinged
that unit with "ff", you'd see the first 4 to 6 characters come bach
trashed, then the rest would be OK.
I'll let you know what happens here.

Chuck Scott, Pres.
Freeway, Inc.
cscott@freeway.net

On Thu, 12 Oct 1995, Dennis Simpson wrote:

> > My PM2ER20 WAN port locks up about every 7 days or so. Didn't do this
> > at first but started about a month ago. There are CRC and FRAME errors
> > but not many (<10). Resetting w1 clears the problem. Not too bad a
> > problem as long as it doesn't go unnoticed for more than a few minutes.
>
> We have seen this as well. We reported it to our carrier, since crc's
> typically indicate line trouble. They called in the telco, who fumbled
> with it for three weeks. They eventually made some line adjustments
> (said the signal was too hot), and we went from outages like you
> describe happening every few hours to outages happening once every
> several days now. They usually are accompanied by a small number of
> crc's like you describe. The telco's test equipment never reports any
> errors, of course. I suppose that if I paid for my CSU/DSU what they
> paid for their test equipment, my CSU/DSU would likely never have
> trouble either. :-)
> The telco swears there is nothing wrong with the line now, and the
> carrier swears we should never get even one crc, since we are on fiber.
>
> So who's not coming clean? I have replaced both the CSU/DSU and the
> router twice now, so I do not believe it is broken hardware. We have a
> new multi-port CSU/DSU coming in in a week or so, and then we will have
> the opportunity to see if the errors go away, in which case it means
> our Black Box CSU/DSU was at fault (compat/config/whatever).
>
> FWIW, we ran without this kind of trouble too, for months. It began
> September 4, when they upgraded our backbone from 128K to 256K. We will
> be going to 512K in the next few weeks, so it should be interesting to
> see what if anything happens to the problem then.
>
> In our case, we almost never reset the port, because it recovers on its
> own in a couple of minutes. Typically by the time we recognize the
> freeze, confirm it and try to reset, it has come recovered. Like you say,
> annoying but not a huge issue.
>
> > Any one else seen this? BTW there are no alarms or errors on my CSU/DSU
> > (ADTRAN TSU).
>
> When these problems occur, the Cisco 7000 at the carrier's end of our
> connection reports crc's as well; our CSU/DSU doesn't have any
> intelligence to be reporting anything to anyone. :-(
>
> > Do I need to upgrade ComOS (currently 3.1.2)? Not just a blanket upgrade
> > statment, please.
>
> Now that you mention it, that is one common factor on every hardware
> combo we have tried. Our 3.1.2 PM2ER-30 otherwise appears to be solid,
> and since horror stories can be heard about any release, we aren't
> going to upgrade ours unless we hear something specific about a fix to
> this particular problem in a new release. We won't upgrade without
> specific reason, since every change carries an element of risk.
>
> I would be interested in any informed feedback on this problem as well.
>
> Thx,
> dennis
>