I am baffled. Tried 3 days. Hope this isn't a silly problem.
I have configured at home a Linux server with 2 Ethernet cards. One is
going to a Cable Modem(eth0), the other to a hub(eth1). The second (client)
computer has one card(eth0), it goes to the same hub.
The server is the gateway. Its address is 192.168.0.1. I configured DHCP
server on eth1, firewall, IP masq, etc. Cable modem uses dhcp too. Here is
my problem. The client computer is a dual boot computer. When I boot
Windows 98 on it, while I look at /var/log/messages on the server, I see
the DHCP request comming in, and the offer and the acceptance of the IP
address, etc by the client. On the client the Windows boot proceeds
normally. I eventually start Internet Explorer and get to the web, see the
cable modem lights blinking as they should, etc. (Don't forget, the cable
modem is on the server eth0 directly, not on the hub). So, I should say
that everything works fine, masquerade, firewall and all.
But when the same client box boots Linux (same Mandrake 7.1 version as the
server), then things don't turn out as they should. Once it brought up its
eth0, it asks for an address. Dhcp server replies, as it did with Windows,
but the client linux box does not seem to have seen it. It waits and tries
again about 10 seconds later. It tries for 60 seconds or so then displays a
DHCP failed.
Never got it to work.
I tought that maybe the irq or io_range where wrong, but I am using an
NE2000 compatible, and the driver is the ne module. The parameters are the
same as under Windows, io_port=0x300 irq=11 probe_pci=1. They are also the
same as on the server box (eth0 is a 3com on the server, eth1 is ne).
Regardless, I tried all the IRQ possible anyway and none helped. If I
change the io_port it does not see the card at all. So I have to conclude
that the values I use are acceptable.
So, I told myself, lets forget about all that DHCP stuff on the client.
I'll assign it the static IP address 192.168.0.62. (On server
/etc/dhcpd.conf set the range from 2 to 60 so I used 62). The netmask I
used were both 255.255.255.0 (first trials) 255.255.0.0 (other trials). I
set the client the gateway to 192.168.0.1. Did everything according to the
book. So, now, when the client comes up, it activates the eth0, ifconfig -a
confirms it, but I can't ping anything but 192.168.0.62 (or loopback). The
server can't ping the client either. I see the lights blink on the hub. It
does not seem that the cable modem gets trafic from those pings (and I
would not expect it to get any).
What is going on! I hate the idea that I got Windows to work with my linux
server while I cant get the same box to work while using linux on it.
BTW: Under the working(sick) windows, ms-prompt box, the ROUTE PRINT
command displays weird routes. Since I am not familiar with that output, I
don't know if it is normal for windows, or not. Maybe the routes are messed
up on the client. I did not activate firewall on it, nor forwarding.
Thanks. All the help is appreciated.
Gerard Perreault