Eth0 Eth1 Strange Problems

zoro

Bloody hell linux is a steamy pile of horse poo but I love. And I hate it all at the same time.

I've just installed a new nic in my server (Redhat 9) so that I could setup traffic shaping and what not on my broadband connection.

The original network card (eth0) was setup on the lan so I didnt bother changing that - I was coing to setup the 3Com (3c905) card on Eth1 as a direct connection to my BB Ethernet Modem.

So, just to test it out before completely losing net connectivity, I slap both nic's into a port on the LAN, giving both their own ip. (192.168.0.2, 192.168.0.3)

Anyway, jsut out of interest, I run "ifconfig eth1 down" as well as "ifdown eth1" (I wanted to test both - just in case)

Lo and behold all connections to my linux box are cut! Wtf?!
So I waddle over to the machine, connect up a monitor and keyboard and undo the damge.
Back to my own PC now and I use a script to take down eth1, sleep for 10 seconds, then bring up both eth nics. This works, but as before all connections are dropped until both nic's are brought back up again!

I try the same, but taking down eth0 this time.
"ifconfig eth0 down"
and all connections to eth1 die, but all connections to eth0 stay alive!

I already found a configuration error that basically said that eth0==eth1==eth0 and fixed it. But now I'm pretty happy that it's all ok!

Here's the output of my ifconfig -a

ifconfig -a
eth0 Link encap:Ethernet HWaddr 00:10:A7:15:1C:00
inet addr:192.168.0.2 Bcast:192.168.0.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:257 errors:0 dropped:0 overruns:0 frame:0
TX packets:97 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:100
RX bytes:23169 (22.6 Kb) TX bytes:11836 (11.5 Kb)
Interrupt:10 Base address:0x3000

eth1 Link encap:Ethernet HWaddr 00:04:75:EF:3B:5D
inet addr:192.168.0.3 Bcast:192.168.0.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:298 errors:0 dropped:0 overruns:1 frame:0
TX packets:59 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:100
RX bytes:30546 (29.8 Kb) TX bytes:4774 (4.6 Kb)
Interrupt:5 Base address:0xb800

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:118 errors:0 dropped:0 overruns:0 frame:0
TX packets:118 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:7580 (7.4 Kb) TX bytes:7580 (7.4 Kb)

as well as my:
/etc/sysconfig/network-scripts/ifcfg-eth0

MTU=""
NETMASK=255.255.255.0
BROADCAST=""
GATEWAY=192.168.0.1
BOOTPROTO=none
IPADDR=192.168.0.2
NETWORK=192.168.0.0
ONBOOT=yes
DEVICE=eth0

and my /etc/sysconfig/network-scripts/ifcfg-eth1

MTU=""
NETMASK=255.255.255.0
BROADCAST=""
GATEWAY=192.168.0.1
BOOTPROTO=none
IPADDR=192.168.0.3
NETWORK=192.168.0.0
ONBOOT=yes
DEVICE=eth1

Also the output of route:

Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
192.168.0.0 * 255.255.255.0 U 0 0 0 eth1
192.168.0.0 * 255.255.255.0 U 0 0 0 eth1
169.254.0.0 * 255.255.0.0 U 0 0 0 eth1
127.0.0.0 * 255.0.0.0 U 0 0 0 lo
default 192.168.0.1 0.0.0.0 UG 0 0 0 eth1

Can anyone help me out here??

Thanks,
Daniel

hostyle

You can't use the same subnet for both NICs and expect routing to work correctly. Try using 192.168.0.2 and 192.168.1.2 instead.

eggshapedfred

whats the output of route when you take eth0 down? it could be that because both interfaces are on the same subnet when you take down one interface the entry for the subnet in the routing table gets removed so any packets going out eth1 don't know where to go.

not sure if this is the problem you're having but if it is just do a

route add -net 192.168.0.1/24 dev eth1 (or eth0)

and you should be back in business.

hth

andy

niallb

default 192.168.0.1 0.0.0.0 UG 0 0 0 eth1

You've just removed your default route, hence no access to anything.
eth0 must have been set up first, then eth1 and the default route got replaced with eth1 as it was the last one activated.

It makes very little sense to run both interfaces on the same network. Make life easy for yourself, and change one.
Keep one NIC for your local network link, and connect
the other one directly to your broadband.
That way you get the chance to filter everything.

NiallB

zoro

Originally posted by niallb
You've just removed your default route, hence no access to anything.
It makes very little sense to run both interfaces on the same network. Make life easy for yourself, and change one.
Keep one NIC for your local network link, and connect
the other one directly to your broadband.
That way you get the chance to filter everything.

That makes alot of sense actually thank you...
I was simply testing the nic's together by putting them both on the same LAN, I just assumed that it would act as a second "door" into the same machine - I wasn't aware that all of this routing trouble would come up.

What I actually intend is for eth1 to be connected directly to my router, so the problem would have disappeared itself had I just done that.

I've just learnt alot about routing though thank you all!

Any more suggestions regarding this setup before I leave the thread to gather dust?

zoro

The automatic routing info that it gave itself (assumed) fecked up my whole eth0/eth1 setup.

And looking back, it's completely right to get fecked up

Anyway, http://www.linuxquestions.org/questions/history/175030 helped me out an awful lot - just in case this thread get's dragged up by some unlucky souls in the future!

Red Alert

Just out of interest, what sort of traffic shaping are you up to?

zoro

Honestly? I'm p*ssed off with my sister grabbing all the bandwidth on her beloved mcFly webboards. So I'm limiting that ip address (of the family pc) to maybe 56K dialup speeds

That way my gaming isn't affected at all

But also I want to to squid proxying and what not to stop her sending mp3's over MSN Messenger. Stupid cow

hostyle

Try upgrading to mySister 2.6.1 - I've heard its much easier to live with.

zoro

Hehe

Ah no, that did sound petty alright though

I just want to stop either machine from grabbing the full bandwidth every time they try to do something - hopefully I understand enough about it now to attempt it. But probably not

halenger

Whinge whinge whine whine.

Use debian.