network broken after kernel upgrade

I screwed up my rootserver. It was running fine with Debian Sarge, kernel
2.4.26. I wanted the 2.6 kernel, so I installed kernel-image-2.6.8-2-868
and module-init-tools. I put the new boot option as default in lilo.conf
and also added the initrd line. Run lilo, all looked fine. After reboot, I
had no access to the server anymore.

No problem, I thought, booted the remote recue system, mounted the
partitions and proc, chrooted into the system and put the old kernel as
default again. Reran lilo, unmounted partitions and rebooted again. This is
where the pain began: still no network access! When I look into kern.log,
it seems that the network driver is loaded ok, but eth0 never gets an ip
address. /etc/network/interfaces is ok too and was not changed in the
process.

I can still reboot into the rescue system. Any ideas?
Eggert

Eggert Ehmke wrote:
> I screwed up my rootserver. It was running fine with Debian Sarge, kernel
> 2.4.26. I wanted the 2.6 kernel, so I installed kernel-image-2.6.8-2-868
> and module-init-tools. I put the new boot option as default in lilo.conf
> and also added the initrd line. Run lilo, all looked fine. After reboot, I
> had no access to the server anymore.
>
> No problem, I thought, booted the remote recue system, mounted the
> partitions and proc, chrooted into the system and put the old kernel as
> default again. Reran lilo, unmounted partitions and rebooted again. This is
> where the pain began: still no network access! When I look into kern.log,
> it seems that the network driver is loaded ok, but eth0 never gets an ip
> address. /etc/network/interfaces is ok too and was not changed in the
> process.
>
> I can still reboot into the rescue system. Any ideas?

If the network interface is supposed to get the setup
by DHCP, have a look at /var/log/daemon.log.

What do following commands report:

ifconfig -a
route -n

Do you have a firewall (ipchains / iptables)? Can
it be blocking the network traffic?

Is DNS set up properly? See /etc/resolv.conf.

HTH

--

Tauno Voipio
tauno voipio (at) iki fi

Tauno Voipio <(E-Mail Removed)> schrieb:

>Eggert Ehmke wrote:
>> I screwed up my rootserver. It was running fine with Debian Sarge, kernel
>> 2.4.26. I wanted the 2.6 kernel, so I installed kernel-image-2.6.8-2-868
>> and module-init-tools. I put the new boot option as default in lilo.conf
>> and also added the initrd line. Run lilo, all looked fine. After reboot, I
>> had no access to the server anymore.
>>
>> No problem, I thought, booted the remote recue system, mounted the
>> partitions and proc, chrooted into the system and put the old kernel as
>> default again. Reran lilo, unmounted partitions and rebooted again. This is
>> where the pain began: still no network access! When I look into kern.log,
>> it seems that the network driver is loaded ok, but eth0 never gets an ip
>> address. /etc/network/interfaces is ok too and was not changed in the
>> process.
>>
>> I can still reboot into the rescue system. Any ideas?
>
>If the network interface is supposed to get the setup
>by DHCP, have a look at /var/log/daemon.log.

I will do so tonight. No access from here.

>What do following commands report:
>
> ifconfig -a
> route -n

when booted in the rescue mode, all looks fine.

>Do you have a firewall (ipchains / iptables)? Can
>it be blocking the network traffic?
>
>Is DNS set up properly? See /etc/resolv.conf.

fw/dns etc was working fine until the reboot, no config changes here.

I suspect something went wrong when I upgraded from woody to sarge.
The upgrade process was ok, but I did no reboot since. So the system
was running some weeks with old woody config and sarge installed.
Eggert

Eggert Ehmke wrote:
> Tauno Voipio <(E-Mail Removed)> schrieb:
>
>
>>Eggert Ehmke wrote:
>>
>>>I screwed up my rootserver. It was running fine with Debian Sarge, kernel
>>>2.4.26. I wanted the 2.6 kernel, so I installed kernel-image-2.6.8-2-868
>>>and module-init-tools. I put the new boot option as default in lilo.conf
>>>and also added the initrd line. Run lilo, all looked fine. After reboot, I
>>>had no access to the server anymore.
>>>
>>>No problem, I thought, booted the remote recue system, mounted the
>>>partitions and proc, chrooted into the system and put the old kernel as
>>>default again. Reran lilo, unmounted partitions and rebooted again. This is
>>>where the pain began: still no network access! When I look into kern.log,
>>>it seems that the network driver is loaded ok, but eth0 never gets an ip
>>>address. /etc/network/interfaces is ok too and was not changed in the
>>>process.
>>>
>>>I can still reboot into the rescue system. Any ideas?
>>
>>If the network interface is supposed to get the setup
>>by DHCP, have a look at /var/log/daemon.log.
>
>
> I will do so tonight. No access from here.
>
>
>>What do following commands report:
>>
>> ifconfig -a
>> route -n
>
>
> when booted in the rescue mode, all looks fine.

Boot with the new system, and check again.

'ifconfig -a' will show if your new kernel has set
up drivers for your network card(s).

>
>>Do you have a firewall (ipchains / iptables)? Can
>>it be blocking the network traffic?
>>
>>Is DNS set up properly? See /etc/resolv.conf.
>
>
> fw/dns etc was working fine until the reboot, no config changes here.

Boot with new system and check again.

> I suspect something went wrong when I upgraded from woody to sarge.
> The upgrade process was ok, but I did no reboot since. So the system
> was running some weeks with old woody config and sarge installed.

Distribution upgrade is a *way* larger operation than
just a kernel change. Nearly all of the distributed
packages were updated (at least in my Debian installations).

Check /etc/network/interfaces for proper setup, it is
the master file for Debian network start-up.

HTH

--

Tauno Voipio
tauno voipio (at) iki fi

Tauno Voipio <(E-Mail Removed)> schrieb:

>Boot with new system and check again.

As I said, it is a remote root server. When booted in normal mode, I
don't have network access. No way. I only can reboot in rescue mode
and inspect logfiles.
Eggert

Eggert Ehmke wrote:

> I screwed up my rootserver. It was running fine with Debian Sarge, kernel
> 2.4.26. I wanted the 2.6 kernel, so I installed kernel-image-2.6.8-2-868
> and module-init-tools. I put the new boot option as default in lilo.conf
> and also added the initrd line. Run lilo, all looked fine. After reboot, I
> had no access to the server anymore.

I changed the IP address from static to DHCP. Now 2.4 kernel is working
again. The 2.6 still gives no IP address to the interface. Ok, will stay
on 2.4 for a while...
Eggert

Eggert Ehmke <eggert.@nickname.berlin.de> writes:

>Eggert Ehmke wrote:

>> I screwed up my rootserver. It was running fine with Debian Sarge, kernel
>> 2.4.26. I wanted the 2.6 kernel, so I installed kernel-image-2.6.8-2-868
>> and module-init-tools. I put the new boot option as default in lilo.conf
>> and also added the initrd line. Run lilo, all looked fine. After reboot, I
>> had no access to the server anymore.

What does this mean. No access how?


>I changed the IP address from static to DHCP. Now 2.4 kernel is working
>again. The 2.6 still gives no IP address to the interface. Ok, will stay

Well, did you tell it to ?


>on 2.4 for a while...
>Eggert