tlug: Dial-up - IMPOSSIBLE!

["Marc E. Christensen" <marc@example.com>]

I am having real problems getting a dial-up into my office working.  The  
weird thing is that it worked fine before upgrading from 1.0 to 1.2.  It
was a clean - start-from-scratch install on clean partitions.

I had slip working on the old system just fine.  I backed up all important
files, re-fdisked, installed and then copied the configurations for dial 
up while reading documentation to make sure I was aware of any changes.
It did not work.  So, I restored the old tools (dip) on the server.  No
changes.  I thought that maybe it was my new 2.0.35 kernel that I had 
compiled which wasn't working.  I restored the old kernel and modules, set
up LILO.  No change when I re-booted.

Go figure.
  
I am at wit's end.

I just can't figure out what is going on.

The dial-up server is on a Class C subnet that seems to be connected to
either a hub or a router.  I can see traffic from other machines than
those on the local cat5 connected machines.

Shouldn't a dial-up server respond to IP traffic intended for it's dialup
IP as if it were that IP?

Once a dial-up connection is established, from home, I can ping, slogin,
telnet, FTP etc. from the dial-up client to the dial-up server.  I cannot,
however, make a connection to any other machine (local or otherwise).
Upon looking at the logs of a machine on the local subnet of the server
diring a ping or other type of connection, I see ICMP redirect requests
coming from the gateway of the subnet.  It's as if the packets go out from
the client, through the dial-up server yet the server does not respond to
anything destined for the dial-up client.

The same machine connects and establishes perfect slip connections with
the school's terminal server.  A similar dip script is used where just the
particulars about logging in have been changed.  I get a class C from this
sever.  So, I know that dial-up works on the client side.

Some questions:
---------------

A.  What possibilities for this behavior exist?

B.  If all machines on the subnet can heat packets from other machines on
    the subnet (non-switched hub) can the gateway do anything to dis-allow
    communication of the dial-up client between other machines on the same
    subnet?

C.  What are the proper routing tables for both client and server for a
    dialup connection like this where the server is on that same subnet as
    the IP it is assigning to it's dial-up client:

        server's Class C:          xxx.yyy.zzz.92
        Clients assigned address:  xxx.yyy.zzz.116

   What should the output of the ifconfig command look like for the sl0
   entries on both client and server?


D.  Could the fact that the Class C addresses are actually subnetted Class
    B addresses (I think).  They start with 128.yyy.zzz.vvv - When I set
    up new machines on this net, the set up scripts always set the default
    subnet mask as 255.255.0.0 which I have to change to 255.255.255.0

Any other ideas?  I can provide the particulars of the routing tables and
ifconfig outputs to specific individuals if that will help.

PS. By the way, I have ppp working as well - exactly the same problems
exist.

Thanks a whole lot!


--
Marc Christensen	       |  Check out the linux-access pages.
SysOp CDSS University of Utah  |  linux-access - Making a difference!
http://ssv1.union.utah.edu     |  http://ssv1.union.utah.edu/linux-access/

---------------------------------------------------------------
Next Meeting: 10 October, 12:30 Tokyo Station Yaesu central gate
Featuring the IMASY Eng. Team on "IPv6 - The Next Generation IP"
Next Nomikai: 20 November, 19:30  Tengu TokyoEkiMae 03-3275-3691
---------------------------------------------------------------
Sponsor: PHT, makers of TurboLinux http://www.pht.co.jp