[c-nsp] Line Code Violations

[John Neiberger]

> I found a similar issue in a local portion of Sprint's network years back.
> It worked mostly fine, except there were certain bit patterns the
> misconfigured circuit just couldn't pass.  That manifested itself as a
> customer complaining that there was a certain binary file on their (web or
> ftp, can't remember) site that many remote users couldn't download.  It'd
> always get to the same point and the xfer would stall.  Testing it
> locally, it worked fine.  That was fun (and educational) to troubleshoot.

I had an issue like that once. The carrier could test the circuit
using 8-bit patterns and they always ran clean. We didn't ever see any
problems except for when we tried to transfer one specific file. The
file transfer would always fail at the same point. I ended up editing
the file in a hex editor and found the precise 18-byte pattern that
would cause it to fail. The carrier couldn't replicate it because
their test equipment couldn't do 18-byte patterns.  :-)

They replaced an old NIU and all was well.

John