On Mon, 2007-01-15 at 17:42 -0500, Nielsen, Steve wrote:

On 2950s the eth devices are assigned their number in the wrong order:

Gb1 becomes eth1 (should be eth0) Gb2 becomes eth0 (should be eth1)

Dell posted software that provides a way for the Linux kernel to name the interfaces correctly.

Here is a link to the white paper: http://linux.dell.com/files/whitepapers/nic-enum-whitepaper-v3.pdf

Steve

-----Original Message----- From: centos-bounces@centos.org [mailto:centos-bounces@centos.org] On Behalf Of Ed Morrison Sent: Monday, January 15, 2007 4:04 PM To: CentOS mailing list Subject: [CentOS] Server Install Eth0 & Eth1 not working

Hi Everyone:

I installed CentOS 4.4 on a new Dell 2950 using the server cd. No errors during the installation and no errors on bootup when loading the Eth0 or Eth1 interfaces. I currently have iptables stopped and selinux disabled. The interfaces have been given a static ip (192.168.1.x). the subnet is 255.255.255.0 and the gateway is 192.168.1.1. I can not ping any device on my network including the gateway. I can ping 127.0.0.1 and the 192.168.1.x. After trying to ping another device ifconfig will show that on ifcfg-lo transmit and receive packets of 389 or some other number depending on how many ping packets were sent. Ifcfg-th0 or Ifcfg-eth1 will not show any TX or RX packets.

The box was ordered to support RH from Dell. The NIC is an embedded dual port Broadcom NetXtreme II5708 GigabitEthernet NIC. I've replaced all network cables and the ports to where they connect into the switch and even a different switch and it still will not show me any love.

There are many issues on the lists.dell.com concerning the broadcom cards. As an example:

http://lists.us.dell.com/pipermail/linux-poweredge/2004-October/thread.html#...

I just checked all my dell servers (I have many) ... but most seem to have intel gigbit NICs.

"Intel Corporation 82541GI/PI Gigabit Ethernet Controller" on most.

Well ... I have this on one machine and it seems to be working OK:

"Broadcom Corporation NetXtreme BCM5721 Gigabit Ethernet PCI Express"

If you use these search terms on google, you will see many issues with broadcom cards on RHEL:

"site:http://lists.us.dell.com/pipermail/linux-poweredge/ tg3"

Nielsen, Steve wrote:

On 2950s the eth devices are assigned their number in the wrong order:

Gb1 becomes eth1 (should be eth0) Gb2 becomes eth0 (should be eth1)

Dell posted software that provides a way for the Linux kernel to name the interfaces correctly.

Here is a link to the white paper: http://linux.dell.com/files/whitepapers/nic-enum-whitepaper-v3.pdf

Steve

This was the issue. Thanks Steve.

Matthew Martz wrote:

On Mon, January 15, 2007 5:03 pm, Ed Morrison wrote:

...

Hi Everyone:

I installed CentOS 4.4 on a new Dell 2950 using the server cd. No errors during the installation and no errors on bootup when loading the Eth0 or Eth1 interfaces. I currently have iptables stopped and selinux disabled. The interfaces have been given a static ip (192.168.1.x). the subnet is 255.255.255.0 and the gateway is 192.168.1.1. I can not ping any device on my network including the gateway. I can ping 127.0.0.1 and the 192.168.1.x. After trying to ping another device ifconfig will show that on ifcfg-lo transmit and receive packets of 389 or some other number depending on how many ping packets were sent. Ifcfg-th0 or Ifcfg-eth1 will not show any TX or RX packets.

The box was ordered to support RH from Dell. The NIC is an embedded dual port Broadcom NetXtreme II5708 GigabitEthernet NIC. I've replaced all network cables and the ports to where they connect into the switch and even a different switch and it still will not show me any love.

Any ideas would be welcome!

Thanks

Ed

Have you set one NIC to not start on boot, reboot and then see if it works?

Test with only one NIC in the equation and see how it turns out.

also check to make sure their hardware addresses are not the same, sometimes as an admin I see folks copy/paste network settings from one card to another in the console via an editor or cp or mv command, and the hardware address gets cat'd in by mistake.

also just use the ifdown and ifup commands and tail -f /var/log/messages as you do so. if that doesn't work you've already seen the driver and other recommendations, -putting in a spare NIC for a server is useful if the testing shows your onboard broadcomms are being iffy. the dells I work on have two broadcomms and one intel -any luck with the 10/100 intel nic on there?

ifdown eth0 ifdown eth1 ifdown eth2 ifup eth0 ifup eth1

you've also seen the mayhem with port#'s not matching actual hardware #'s, so maybe skip to watching the reboot dmesg output by doing

dmesg | less and scanning for the hardware

also do:

lspci

that should show you those cards sitting on the board, if not, then you have a firmware/bios/hardware driver issue. try updating to another newer kernel see if that helps. get the dell tool johnny hughes mentioned, or even go to broadcomm and see if there's a linux driver -perhaps the broadcomm nic type your dell has only works with wintel -my own vendors have slipped before when telling me a server is linux ready -only to find out they didn't vet this info with their hardware/assembly dept.

if that turns out to be the case, you should catapult cows at them.

-karlski

Is this not the time to make use of the HWADDR option in your /etc/sysconfig/network-scripts/ifcfg-eth* files?

http://www.centos.org/docs/4/html/rhel-rg-en-4/s1-networkscripts-interfa ces.html

"HWADDR=<MAC-address>, where <MAC-address> is the hardware address of the Ethernet device in the form AA:BB:CC:DD:EE:FF. This directive is useful for machines with multiple NICs to ensure that the interfaces are assigned the correct device names regardless of the configured load order for each NIC's module. This directive should not be used in conjunction with MACADDR. "

Hope this helps,

J

-----Original Message----- From: centos-bounces@centos.org [mailto:centos-bounces@centos.org] On Behalf Of Ed Morrison Sent: January 15, 2007 8:48 PM To: CentOS mailing list Subject: Re: [CentOS] Server Install Eth0 & Eth1 not working

...

also check to make sure their hardware addresses are not the same, sometimes as an admin I see folks copy/paste network settings from one

card to another in the console via an editor or cp or mv command, and the hardware address gets cat'd in by mistake.

also just use the ifdown and ifup commands and tail -f /var/log/messages as you do so. if that doesn't work you've already seen the driver and other recommendations, -putting in a spare NIC for

a server is useful if the testing shows your onboard broadcomms are being iffy. the dells I work on have two broadcomms and one intel -any luck with the 10/100 intel nic on there?

ifdown eth0 ifdown eth1 ifdown eth2 ifup eth0 ifup eth1

you've also seen the mayhem with port#'s not matching actual hardware #'s, so maybe skip to watching the reboot dmesg output by doing

dmesg | less and scanning for the hardware

also do:

lspci

that should show you those cards sitting on the board, if not, then you have a firmware/bios/hardware driver issue. try updating to another newer kernel see if that helps. get the dell tool johnny hughes mentioned, or even go to broadcomm and see if there's a linux driver -perhaps the broadcomm nic type your dell has only works with wintel -my own vendors have slipped before when telling me a server is

linux ready -only to find out they didn't vet this info with their hardware/assembly dept.

if that turns out to be the case, you should catapult cows at them.

-karlski

Thanks for the reply Karlski. Steve had the issue nailed in his post with the NIC's being named wrong. That said there appears to be ongoing weirdness with RH and Broadcoms in general. Replacing seems to be the prudent measure. _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos