• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

fedora(redhat) networking + multiple interfaces = no joy

spyordie007

Diamond Member
I have another problem that I'm hoping you guys can help me out with, I've got 2 network interfaces on my laptop and I cant get them to play nice together. I have an onboard 802.11b adaptor (Prism2; eth1) and a PCMCIA ethernet (3CFE575; eth0) adaptor. I'm using redhat-config-network 1.3.10 for network configuration. With both adaptors logically installed eth0 will not start on boot and does not work however eth1 works fine, if I uninstalled eth1 (logically or by disabling it in the BIOS) than eth0 starts as normal and works however as soon as I reinstall eth1 eth0 fails to start on boot.

Does anyone have any suggestions as to how I could go about troubleshooting this?

One other thing, I noticed that bash doesnt have the ifconfig command. How do I get it "installed"

Thx

-Spy
 
I should also add that when eth0 does not come up on the boot the system will not shut down properly, it hangs on "routing card interrupts to PCI" :|
 
am i imagining it or do you always post twice in a row? not to flame or anything....

but there is an "edit" button in case you missed it
 
to run ifconfig, you need to be root, and have /sbin in your path. The easiest way is to make sure you change to root using su -
 
Originally posted by: spyordie007
well that would explain why ifconfig isnt working, what is the syntax for adding the /sbin path to my bash profile?

If it looks something like this:

PATH=$PATH:$HOME/bin

then change it to

PATH=$PATH:$HOME/bin:/sbin

 
shouldn't have a problem having 2 interfaces, actually you shouldn't have a problem running 8 of them.

ifconfig isn't normally in the path of a user, only root should be able to access it, plus just switching to "su" won't set the path properly, but that's not a big deal to add /sbin to your stuff. 🙂 I actually prefer to use "sudo su" then that sets all the eviromental stuff automaticly, but you have to allow your user to use sudo thru editing the /etc/sudo.users file.

Could be a resource confict like both devices running irq 5 or something. Try to select "using a non-plug'n'play OS" stuff in your bios. Or maybe if it already selected as a non plugnplay OS, select the other one. Also maybe try (just for a test) turning off usb support or sound card suppor or serial ports or everything and see if that makes a difference. Modern BIOS should handle the irq/mem address assignment nowadays, but sometimes the don't work all that hot and the OS has to, this plugnplay aware OS controls that in a course way.

Could try turning off acpi support, but since this is a laptop that may not be so hot. Pass acpi=off to the kernel at boot up.

Also check out the "dmesg" command output.

If you can't get it to work so that you can have both Eth0 devices going at once you can try
ifconfig eth1 down
ifconfig eth0 up
commands, to switch back and forth.

Then if you can switch back and forth easily like that, then you can put that into a script (and use a simple lock file)

#!/bin/bash
# location: /sbin/ name: switch.sh

if grep 1 ~/var/lock/ifconfig.lock
then
ifconfig eth1 down
ifconfig eth0 up
echo "0" > ~/var/lock/ifconfig.lock
else
ifconfig eth0 down
ifconfig eth1 up
echo "1" > ~/bin/xawtv.lock
fi

Then run it (if you have sudo setup) like thus:
sudo /sbin/switch
to get the one interface working, then:
sudo /sbin/switch
again to get the other one.


Technicly you may have to make the script more complicated or not. The way to get interfaces working correctly while doing manually would be:

modprobe driver.for.device
ifconfig eth0 192.168.1.10 up
route add default gw 192.168.1.254

Then to take it down:

route del default gw
ifconfig eth0 down
rmmod driver.for.device

or at least something close to that.

If you want to get fancy then you can make a script for a bunch of different networks:
#!/bin/bash
# location: /sbin name: switch syntax: /sbin/switch network.name

if [ $1 == "home" ]
then
echo "setting up home"
#here it sets up a bunch of stuff for that network
fi

if [ $1 == "roaming" ]
then
echo "setting up wireless"
#here it sets up a bunch of stuff for that network
fi

if [ $1 == "work" ]
then
echo "setting up work"
#here it sets up a bunch of stuff for that network
fi

if [ $1 == "hacked.corporate.lan" ]
then
echo "setting up hacked.corporate.lan"
#here it sets up a bunch of stuff for that network
fi


Then you would just go:
sudo /sbin/switch home
for your home
or maybe
sudo /sbin/switch work
to switch to the network lan.

Also to use dhcp you can

dhcpcd eth0
or maybe
dhclient eth0

or something like that.
 
I just had a similar problem with redhat 9. I installed redhat on a machine with one ethernet card, and everything was fine. Then I put in another ethernet card, and redhat didn't automatically detect it. So I ran kudzu manually and it detected it, but it still didn't bring the interface up. After i brought the interface up ifup eth1, neither interface worked. I needed to take both down, then bring up only eth1.

I haven't figured out the problem yet, will try again on monday night. But, I'll be watching this thread for ideas.
 
Funky irrelevant script tricks aside, it should still allow 100% operation of both NICs simultaneously. To me, it sounds like both NICs might somehow be sharing an interrupt (hence why it might hang on "routing card interrupts to PCI").

Another thing you might want to check is to see if there are any noted problems with your model of laptop. It's possible that the drivers for the PCMCIA controller or the wireless NIC aren't up to par, which may cause the hang you mentioned. Once you can unequivocally rule that out, start looking at the h/w s/w combination. Does the same setup work fine with slack (as others may or may not have reported somewhere), etc..

I'd post the question to the fedora list, since it's likely someone else either has or will have this problem. 🙂
 
Originally posted by: chsh1ca
Funky irrelevant script tricks aside, it should still allow 100% operation of both NICs simultaneously. To me, it sounds like both NICs might somehow be sharing an interrupt (hence why it might hang on "routing card interrupts to PCI").

Funky irrelivent scripts? Why not if they work? Your probably not going to use both at the same time so why not take the easy way out?

If something like:
Could be a resource confict like both devices running irq 5 or something. Try to select "using a non-plug'n'play OS" stuff in your bios. Or maybe if it already selected as a non plugnplay OS, select the other one. Also maybe try (just for a test) turning off usb support or sound card suppor or serial ports or everything and see if that makes a difference. Modern BIOS should handle the irq/mem address assignment nowadays, but sometimes the don't work all that hot and the OS has to, this plugnplay aware OS controls that in a course way.

doesn't work then scripts may be the only way to "fix" it until they get better updates.
 
Originally posted by: drag
Originally posted by: chsh1ca
Funky irrelevant script tricks aside, it should still allow 100% operation of both NICs simultaneously. To me, it sounds like both NICs might somehow be sharing an interrupt (hence why it might hang on "routing card interrupts to PCI").

Funky irrelivent scripts? Why not if they work? Your probably not going to use both at the same time so why not take the easy way out?
I didn't mean to offend, I was just saying that it is irrelevant to the problem. It may be a potential solution, but it is not a good one in that this is something that shouldn't be an issue. There shouldn't be any reason it won't work with both interfaces configured and up. IMO, settling for turning off one interface while primarily using another is a littly silly, especially on an operating system designed to be networked.
 
well it looks like it was indeed an IRQ issue that was causing the problem, after playing around with IRQs on the interfaces I finally found a configuration which will allow both to be up and running at the same time. I've also added sbin to my user path so that I can run basic commands from the prompt.

However the problem has not been completely resolved as DNS is not working as it should be, DNS appears to only be resolving addresses based upon the DNS server recieved (via DHCP) by the last interface to come up. My interfaces are on 2 differant networks and when I try and use host names on each network the only way I can get it to fuction is to bring that interface down/up. Manually specifying my DNS server(s) in redhat-config-network (1.3.10) gets written over whenever a dhcp lease gets renewed.

Manually specifying my DNS servers would be okay (if someone can tell me a way that will keep the settings); however if anyone knows how I can get this working the "right" way I would appreciate it.

Thanks again,

-Spy
 
That one's easy.

it's your /etc/resolv.conf file.

Just put:

nameserver dns.ip.address

If the file isn't there, just create one.

You can have as many dns server ip addresses in there as you'd like. They search in order.
 
my dns servers and search domain are already listed there; it says that it was generated by /sbin/dhclient-script

if this is something that is generated for me than are my problems because it is getting overwritten?

-Spy
 
Are you sure it's getting overwritten? Try putting a comment or changing the order of DNS servers and see if it gets changed because of the script...

Otherwise can you ping out? Can you ping 127.0.0.1, your own ip address, the gateway address, google's address (216.239.41.99), etc etc?

If it can't reach the DNS address then it can't resolve the dns name... If you take down one of the eth* interfaces, does that work then?
 
I had teh same problem with Red Hat 8 and Red Hat 9 and now Fedora, but i did find one easy fix. Configure all your network interfaces during installation. That way eth0 is assigned to eth0 and eth1 is assigned to eth1 and so on. Also don't change r network setting using the Red Hat control applet because it will cause you problems again.
 
Back
Top