Re: gateway changed: can't get out
From: P Gentry (rdgentry1_at_cablelynx.com)
Date: 14 Mar 2004 20:56:26 -0800
Dave Arnette <firstname.lastname@example.org> wrote in message news:<gv55c.7332$JL2.146006@attbi_s03>...
> /dev/rob0 wrote:
> > On Sun, 14 Mar 2004 19:58:31 +0000, Dave Arnette wrote:
> >> The IP address of the cable modem router: 192.168.99.1
> >> The IP address of my linux box: 192.168.1.11
> > Why didn't you change the IP of the Linux box too?
> >> What I have done to try and resolve the problemis:
> >> 1) Created a new interface on the ethernet card with the IP address of
> >>192.168.99.11 and called it eth0:1 as follows:
> >>eth0:1 Link encap:Ethernet HWaddr 00:A0:24:75:B1:94
> >> inet addr:192.168.99.11 Bcast:192.168.99.255 Mask:255.255.255.0
> >> 2) created a 'route' to the 192.168.99.xxx network by doing the following:
> >> route add -net 192.168.99.0 netmask 255.255.255.0 dev eth0:1
> >>After the route command, I can then ping varions other boxes on the
> >>192.168.99 network, like the cable modem route itself:
> > So you did it the hard way and made it work ...
> >>But I can't ping anything outside of my house.
> > ... but you don't know how to configure your router. If it's a Linux
> > machine perhaps you can ask here. Otherwise call the router's tech
> > support number.
> I could be convinced that the problem lies on the cable modem router,
> except the Sun machines are working just fine with a new interface
> (192.168.99.xx) on the ethernet and a new 'route'. Even the windows
> machines are working fine that way. It's only the Linux machines that
> are broken, so I don't think it anything todo with the cable modem router.
Holding back vital info/clues, eh? Actually, comparing routing tables
from a working eg. and the brokeRH one would help. Ditto for any
router between broke and CM. Route -C after some activity (as well as
arp) would also help. Look at the output and see if you spot
something amiss with brokeRH compared to working eg.
> There's got to be an easier way though. 192.168.99.xx should work, but
> it's not.
> Thanks Dave
Why you think having two nets on one segment would work reliably and
automatically? Especially when the two net "devices" have the same
MAC? Which one is making arp requests by broadcasting, which one by
routing to the gw -- ie., which segment/IP is local?
Are you sure the routing tables are set up correctly on _all_ hosts on
the network -- ie., the brokeRH and the router it's using as gw? The
routing tables will have to "fully" support both nets on the brokeRH
interface just as if it had two nics, each on a different net. Do
they? Ditto for the interface feeding brokeRH.
Are you running any DHCP hosts on the LAN or are they all static and
running through a NAT?
Do any of the Solaris or Win boxes have aliased nics? Do they work?
How does their config compare with brokeRH?
BTW, is the CM separate from the router or is this a combined
Tend to agree with rob0 that you seem to have made this more
complicated than need be by just throwing another IP/net segment into
the mix of an existing setup -- it's usually not that easy.
If your "other" hosts are working OK, I would suspect a faulty routing
table or two and pick them apart. Then look keenly at NAT rules after
the route tables are confirmed OK -- ie., get the LAN side working
fully, then work out to the internet devices. Be especially mindful
of "return" traffic that will have to be properly NATed to brokeRH.
email above disabled