Debian: eth0 loses network configuration

Hi folks,

I'm running a Sarge system on a Commell LV-671 board with an
Intel-82541PI-Gigabit-NIC. I configured it in /etc/network/interfaces
like this:

# The primary network interface
auto eth0
iface eth0 inet static
address 10.200.40.2
netmask 255.255.255.0
broadcast 10.200.40.255
network 10.200.40.0
gateway 10.1.0.254


In my opinion, the gateway-line should be commented out, but I was
overuled. Roumors say, it didn't change anything, I haven't tried and
seen it myself yet, though.

When entering runlevel 2, a script is called, setting a default gateway
to be able to connect to the system from the 10.1.0.0 network:

/sbin/route add -host 10.1.0.254 eth0
/sbin/route add default gw 10.1.0.254

(only relevant parts - the 10.200.0.0 network is not able to connect to
the internet)

This seems to work fine. In our network, there is a DHCP server. After a
certain time, eth0 "drops" its network address and plays around with the
DHCP server. Say: It fetches an IP address and reconfigures. The only
thing conspicious I could find, is the following message in the kernel
logs:

Mar 14 09:17:02 localhost kernel: e1000: eth0: e1000_watchdog: NIC Link
is Up 1000 Mbps Full Duplex

It pops up from time to time, without claiming the link having been down
before. So I can find this message several times in a row in the kernel
log. Every time this message appears, I have to face the described
problem. It find it quite interesting, that one time, it works for
hours, even days, but some other time, only for ten minutes. Then maybe
again for two hours... So there are always different intervals.

/etc/init.d/networking restart (and executing the routing script)
works for the time described above. It says "network unreachable",
though, but I'm quite sure that this is caused by the unknown gateway
(10.1.0.254). I assume that this is also the reason for the "failed to
bring up eth0" message, that follows right after the "network
unreachable" line?


Any ideas?

Muffi

On Wed, 14 Mar 2007, Björn Marschollek wrote:

> Hi folks,
>
> I'm running a Sarge system on a Commell LV-671 board with an
> Intel-82541PI-Gigabit-NIC. I configured it in /etc/network/interfaces
> like this:
>
> # The primary network interface
> auto eth0
> iface eth0 inet static
> address 10.200.40.2
> netmask 255.255.255.0
> broadcast 10.200.40.255
> network 10.200.40.0
> gateway 10.1.0.254
>
>
> In my opinion, the gateway-line should be commented out, but I was
> overuled. Roumors say, it didn't change anything, I haven't tried and
> seen it myself yet, though.
>
> When entering runlevel 2, a script is called, setting a default gateway
> to be able to connect to the system from the 10.1.0.0 network:
>
> /sbin/route add -host 10.1.0.254 eth0
> /sbin/route add default gw 10.1.0.254
>
> (only relevant parts - the 10.200.0.0 network is not able to connect to
> the internet)
>
> This seems to work fine. In our network, there is a DHCP server. After a
> certain time, eth0 "drops" its network address and plays around with the
> DHCP server. Say: It fetches an IP address and reconfigures. The only
> thing conspicious I could find, is the following message in the kernel
> logs:
>
> Mar 14 09:17:02 localhost kernel: e1000: eth0: e1000_watchdog: NIC Link
> is Up 1000 Mbps Full Duplex
>
> It pops up from time to time, without claiming the link having been down
> before. So I can find this message several times in a row in the kernel
> log. Every time this message appears, I have to face the described
> problem. It find it quite interesting, that one time, it works for
> hours, even days, but some other time, only for ten minutes. Then maybe
> again for two hours... So there are always different intervals.
>
> /etc/init.d/networking restart (and executing the routing script)
> works for the time described above. It says "network unreachable",
> though, but I'm quite sure that this is caused by the unknown gateway
> (10.1.0.254). I assume that this is also the reason for the "failed to
> bring up eth0" message, that follows right after the "network
> unreachable" line?
>
>
> Any ideas?

It sounds like a driver issue. Try using a different or newer version
of the driver.

--
Sir Jackery

> It sounds like a driver issue. Try using a different or newer version of
> the driver.

Changing the motherboard fixed it. The same problem occured on other
operation systems, too...