Came across something similar a few months ago, if you have a mixed Linux
and Windows Network and you are running Samba and NAT (are you?) then
unpredictable problems arise from the Linux end. Apparently documented in
the Microsoft Tech Net. This particular box was a Cobalt Cube which was
reverted to a previous version of the software that killed the problem.
Karl
----- Original Message -----
From: "Paul Borghese"
To:
Sent: Sunday, June 03, 2001 2:55 AM
Subject: RE: Problem with home network [7:6922]
> Yea, I have a theory. It has to do with the Linux box, defaulting to
> 100MB. Let's suppose the Linux box and PC are both running at 100 MB/sec
> but the hub is only 10 MB/sec. Maybe the timing is such that it will not
> work past one hop.
>
> When I recompiled the kernel, maybe the new kernel release changed how the
> modules works on the Ethernet card, causing it to not detect 10 MB/sec
> connections and to default to 100 MB/sec.
>
> I will do a diff on the code.
>
>
> Who knows?
>
> Paul
>
> Daniel Cotts wrote:
> >
> > Since you touched the Linux box it would be the first suspect.
> > Can you verify that there was no configuration change? - even
> > by accident?
> > If there was a change, can you roll it back to original?
> > Are there other computers or printers connected to the hub?
> > Is the hub single speed or dual speed? (10/100)
> > (Thinking about speed/duplex mismatches.)
> > How does the Linux box configure the default route? Does it
> > point to its own
> > E0 interface or to the remote GW? (Thinking about filling its
> > ARP cache)
> >
> > > -----Original Message-----
> > > From: Paul Borghese [mailto:[EMAIL PROTECTED]]
> > > Sent: Saturday, June 02, 2001 7:42 PM
> > > To: [EMAIL PROTECTED]
> > > Subject: Problem with home network [7:6922]
> > >
> > >
> > > I have a cable modem connected to a linux box that is
> > > performing NAT from my
> > > invalid home network of 172.16.1.0/24 to my valid IP address
> > > 209.160.20.67.
> > > The 172.16.1.0 network is going into a small inexpensive hub.
> > > This setup
> > > has worked for about a year.
> > >
> > > A few days ago, I needed to do some things on the linux box.
> > > When I hooked
> > > everything back up my internet access was horrid. Found the
> > > following:
> > >
> > > If I ping from 172.16.1.98 (my PC) to the following addresses:
> > >
> > > 172.16.1.1 (PC's Default GW, E1 interface on Linux box)
> > > 0% Packet Loss
> > > 209.160.20.67 (E0 Linux IP address and address PC is being
> > > NATed to) 0%
> > > Packet loss
> > > 209.160.20.1 (GW of Linux Box)
> > > 70% Packet Loss
> > >
> > > If I ping from the Linux box I see no packet loss to
> > 172.16.1.98 or
> > > 209.160.20.1. So I can now deduce the connection between the
> > > Linux box and
> > > the default GW is clean.
> > >
> > > But something is occuring with the NAT translations that
> > > causes 70% packet
> > > loss through the box.
> > >
> > >
> > > Ok, so here is the puzzling thing. If I remove the hub and
> > > use a crossover
> > > cable between the PC and Linux box the address which is
> > problamatic
> > > 209.168.20.1 receives no packet loss when pinging from the PC
> > > - hence fixing
> > > the problem.
> > >
> > > So in other words, removing the hub on the 172.16.1.0 network
> > > fixes the
> > > connection at 209.168.20.1 ?!?
> > >
> > > Any ideas?
> > >
> > >
> > > Paul Borghese
> > > Report misconduct
> > > and Nondisclosure violations to [EMAIL PROTECTED]
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=6968&t=6922
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
