Odd networking problem.

Symptom:
Only one of the interfaces can contact IP adresses outside of the local
machine.

Enable eth1 means eth0 stops functioning. Disable eth1 means eth0
starts working again.
Enable eth0 means eth1 stops functioning. Disable eth0 means eth1
starts working again.



ping 217.75.113.33 and ping 217.75.113.34 works fine from the local
machine, so the NIC's doesnt have any inherent errors within themselfs.


System:
Linux sokaris 2.6.8-11-em64t-p4-smp #1 SMP Mon May 30 22:45:17 UTC 2005
x86_64 GNU/Linux
Hardware: Mainboard (ASUS NCCH-DL I875P), Dual 3.2GHZ Xeon CPUs, 4GB
DDR RAM, 10k RPM SATA disk.
Network chip: Intel LAN S82547GI, Intel LAN S82540EM)
Network driver: e1000 (loaded as module)

The following is output from /etc/network/interfaces, ifconfig, route
and ping symptoms.

======================
sokaris:/home/global# cat /etc/network/interfaces

auto lo
iface lo inet loopback

# The primary network interface
auto eth0
iface eth0 inet static
address 217.75.113.33
netmask 255.255.255.0
gateway 217.75.113.1


auto eth1
iface eth1 inet static
address 217.75.113.34
netmask 255.255.255.0
gateway 217.75.113.1
======================

sokaris:/home/global# route
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use
Iface
217.75.113.0 * 255.255.255.0 U 0 0 0
eth0
217.75.113.0 * 255.255.255.0 U 0 0 0
eth1
default cr1-2-8.shg5.g. 0.0.0.0 UG 0 0 0
eth1
default cr1-2-8.shg5.g. 0.0.0.0 UG 0 0 0
eth0

======================

Ping from eth0 to local router works fine
sokaris:/home/global# ping -I eth0 217.75.113.1
PING 217.75.113.1 (217.75.113.1) from 217.75.113.33 : 56(84) bytes of
data.
64 bytes from 217.75.113.1: icmp_seq=1 ttl=255 time=0.266 ms
64 bytes from 217.75.113.1: icmp_seq=2 ttl=255 time=0.257 ms

--- 217.75.113.1 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 999ms
rtt min/avg/max/mdev = 0.257/0.261/0.266/0.016 ms


Ping from eth1 to local router doesn't work. It does work If I disable
eth0.
sokaris:/home/global# ping -I eth1 217.75.113.1
PING 217.75.113.1 (217.75.113.1) from 217.75.113.34 : 56(84) bytes of
data.
--- 217.75.113.1 ping statistics ---
4 packets transmitted, 0 received, 100% packet loss, time 2998ms


======================


sokaris:/home/global# ifconfig
eth0 Link encap:Ethernet HWaddr 00:118:8D:05:0D
inet addr:217.75.113.33 Bcast:217.75.113.255
Mask:255.255.255.0
inet6 addr: fe80::211:d8ff:fe8d:50d/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:195168 errors:0 dropped:0 overruns:0 frame:0
TX packets:1145 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:12963984 (12.3 MiB) TX bytes:138545 (135.2 KiB)
Base address:0xa000 Memory:f1020000-f1040000

eth1 Link encap:Ethernet HWaddr 00:118:8D:05:0E
inet addr:217.75.113.34 Bcast:217.75.113.255
Mask:255.255.255.0
inet6 addr: fe80::211:d8ff:fe8d:50e/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:2183231 errors:0 dropped:0 overruns:0 frame:0
TX packets:2750254 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:828543825 (790.1 MiB) TX bytes:1681281277 (1.5 GiB)
Base address:0xb400 Memory:f4000000-f4020000

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:12 errors:0 dropped:0 overruns:0 frame:0
TX packets:12 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:1008 (1008.0 b) TX bytes:1008 (1008.0 b)

On 14 Aug 2005 07:17:44 -0700, (E-Mail Removed) wrote:
> Symptom:
> Only one of the interfaces can contact IP adresses outside of the local
> machine.

You should have only one nic set as the default gateway and the
interfaces should be on seperate networks.

The G in the Flags column indicates the default gateway.
$ route -n
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
192.168.2.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0
24.1.202.0 0.0.0.0 255.255.254.0 U 0 0 0 eth1
0.0.0.0 24.1.202.1 0.0.0.0 UG 0 0 0 eth1


For hardware layout/settings Mandrakelinux example:

http://groups.google.com/advanced_group_search
(E-Mail Removed)lid in the Message Id box.

So it is not possible to have two NIC's in the same network?

..33 and .34 are the same machine on the same network, and they should
use the same gateway.

Should I only set the gateway-option (in /etc/network/interfaces) once,
and does it matter WHERE I put it?

On 14 Aug 2005 07:46:17 -0700, (E-Mail Removed) wrote:


You should leave enough of the article's verbage so everyone can tell who you
are responding to.


> So it is not possible to have two NIC's in the same network?

No idea. Normally you have one gateway interface and any traffic that will
not route to other interfaces will be sent to the gateway interface.

> .33 and .34 are the same machine on the same network, and they should
> use the same gateway.
>
> Should I only set the gateway-option (in /etc/network/interfaces) once,

Please read http://www.catb.org/~esr/faqs/smart-questions.html
You have to provide what distribution and version for anyone
to provide exact information.

> and does it matter WHERE I put it?

Usually, keyword=value order does not matter in config files and in
the case of duplicates, last one supercedes previous values in the
same config file.

I have read Eric S. Raymonds "smart questions" page.

The distribution is Debian 3.1 unstable with the 2.6.8-11-em64t-p4-smp
vanilla kernel (installed with apt-get).

However, I'm sorry if anything was unclear. Working with a problem for
a couple of hours makes you think everyone knows about the problem when
you post it....and for that, I am sorry.

Currently, ping -I eth0 www.sunet.se for instance gives me
>From iface01 (217.75.113.33) icmp_seq=1 Destination Host Unreachable

While the same from eth1 works like a charm.


Do you think it will work fine, if I just delete one of the
gateway-entries in /etc/network/interfaces ?

Thanks in advance

On 14 Aug 2005 08:16:49 -0700, (E-Mail Removed) wrote:
> The distribution is Debian 3.1 unstable with the 2.6.8-11-em64t-p4-smp
> vanilla kernel (installed with apt-get).

Well, that cuts me out of the loop. I run Mandrivalinux

> However, I'm sorry if anything was unclear. Working with a problem for
> a couple of hours makes you think everyone knows about the problem when
> you post it....and for that, I am sorry.

No problem. Been there done that.

>
> Currently, ping -I eth0 www.sunet.se for instance gives me
>>From iface01 (217.75.113.33) icmp_seq=1 Destination Host Unreachable

If ping -c 1 192.36.125.18 fails, it is a routing/firewall problem.
If works then it is a DNS problem.
/etc/resolv.conf contains dns server ip addresses.

> While the same from eth1 works like a charm.
>
>
> Do you think it will work fine, if I just delete one of the
> gateway-entries in /etc/network/interfaces ?

Yes.

On 2005-08-14, (E-Mail Removed) <(E-Mail Removed)> wrote:

> The distribution is Debian 3.1 unstable with the 2.6.8-11-em64t-p4-smp
> vanilla kernel (installed with apt-get).

No, it isn't. Debian 3.1 is Stable.

http://www.us.debian.org/releases/stable/
--
Carl Fink (E-Mail Removed)
If you attempt to fix something that isn't broken, it will be.
-Bruce Tognazzini

Yeh, sorry..been calling it unstable for so long.... I knooow, its
stable

Anyway.... the most frustrating thing is that I know it CAN work,
because I fiddled around for a bit last time.. and wham, it worked...
then I had to reboot, and now this is the problem Im facing... and I
dont remember what I did to make it work last time

On 14 Aug 2005 08:27:28 -0700, (E-Mail Removed) wrote:

> Anyway.... the most frustrating thing is that I know it CAN work,
> because I fiddled around for a bit last time.. and wham, it worked...

Hey, open that righthand desk drawer, pull out and read your system admin diary
to see what you did when, how.

> then I had to reboot, and now this is the problem Im facing... and I

Hmmm, usually that indicates, you did not modiy config files and probably
used route add/del to get the routing straighened out.

> dont remember what I did to make it work last time

Yeah, thats what I'm thinking.. problem is, I dont have physical access
to the machine, so if I screw up and need to reboot, I'll have to wait
for the on-site techs to show up (its sunday) and wait for them to do
it.

Anyway... I'll get this sorted... sometime

Thanks for your help though, much appreciated!