Ascend and TTL

Sat Jul 26 14:07:09 UTC 1997

I apologize for not including any business or philosophical perspectives on
either DNS or fiber management, but I have just learned something about the
little Ascend pipeline boxes that some of our clients use which bothers me.
I was wondering if anyone has run across this one before. I checked their
web site for tech support, but the goofs don't even provide a search func
for technical information. The nanog archive at Merit returns nothing useful
for this topic when searched by me.

Why do packets going through the pipeline products have their ttl chopped
drastically?

For example:

traceroute to pc1.va.connex.net (205.210.187.50): 1-30 hops, 38 byte packets
 1  pro1.barrie.connex.net (205.189.201.1)  1.10 ms  0.795 ms  0.753 ms
 2  p25.va.connex.net (205.210.187.49)  28.2 ms (ttl=63!)  27.6 ms (ttl=63!)  
67.7 ms (ttl=63!)
 3  pc1.va.connex.net (205.210.187.50)  28.7 ms (ttl=30!)  37.6 ms (ttl=30!)  
28.9 ms (ttl=30!)

The pc on this user's site is a typical win95 box. pro1 happens to be a
livingston pm3, but we see this with cisco 1005, 2501 as well. The drop
always happens at the Ascend. Another customer has a connection to a
different provider and his provider uses an Ascend to connect to my client's
ascend, and both routers in the path do this, so his reach onto the net
appears to have a very limited distance (hop-wise). Code revs in the
pipelines is 4.6C in one case, 4.5patch something in the other. Both
exhibit similar behaviour.

I don't use Ascend's and have no manuals (learned a long time ago they
give me heartburn), so I was hoping someone on this list would be familiar
with this behaviour and might be able to suggest how to correct it.

Any takers?

Thx,
dennis