broken e1000 network interface?

Issues related to hardware problems
Post Reply
dakster
Posts: 1
Joined: 2006/06/08 18:50:03

broken e1000 network interface?

Post by dakster » 2006/06/08 19:10:34

I just upgraded one of our computers at work that's been chugging along on redhat 8.0 of all things for ages to centos 4.3, and I'm having some hardware issues. This machine has a supermicro X5DA8 motherboard with dual xeons, an onboard gigabit card, and an additional gigabit card we've put in it (it's on two network simultaneously). Both worked fine in redhat 8.0, but with both the stock kernel from the 4.3 install, and now with a vanilla kernel (2.6.16.20) one of the two intel gigabit's (the onboard one) no longer works. Here's an example of what it does when I bring it up, in either kernel:

[root@opus ~]# ifconfig
eth0 Link encap:Ethernet HWaddr 00:30:48:70:C2:21
inet addr:192.168.0.110 Bcast:192.168.0.255 Mask:255.255.255.0
inet6 addr: fe80::230:48ff:fe70:c221/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:4294966948 errors:4294965556 dropped:4294966948 overruns:0 frame:4294966252
TX packets:4294966948 errors:4294966600 dropped:0 overruns:0 carrier:4294966252
collisions:4294966948 txqueuelen:1000
RX bytes:4294966948 (3.9 GiB) TX bytes:4294966948 (3.9 GiB)
Base address:0x3000 Memory:c3120000-c3140000

eth1 Link encap:Ethernet HWaddr 00:0E:0C:33:2C:00
inet addr:XXX.XXX.XXX.XXX Bcast:XXX.XXX.XXX.XXX Mask:255.255.XXX.XXX
inet6 addr: fe80::20e:cff:fe33:2c00/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:75744 errors:0 dropped:0 overruns:0 frame:0
TX packets:1096 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:26491182 (25.2 MiB) TX bytes:210903 (205.9 KiB)
Base address:0x4000 Memory:c6020000-c6040000


eth1 is the working on that's an addon card, eth0 is the onboard with tons of errors. That's immediately after bringing it up. I'm full up on slots so I can't just drop another one in there. Plus, this worked fine before centos went on there, so I have a feeling it's indicative of bigger problems. It doesn't matter if there's a cable plugged into this ethernet port or not, so it's not a bad cable or switch. Anyone have any tips or clues here? Thanks a ton in advance!

doug8429
Posts: 10
Joined: 2006/08/02 21:27:06
Location: Portland, Oregon

Re: broken e1000 network interface?

Post by doug8429 » 2006/08/03 01:07:48

I can't help you with your problem, but I have the similar issue. I just got a new Supermicro box with dual Xeons and
dual onboard ethernet adapters. The install does not detect them and nothing I've tried has resolved the problem.

Did you ever figure it out?

teleport
Posts: 114
Joined: 2005/09/09 05:30:18
Contact:

Re: broken e1000 network interface?

Post by teleport » 2006/08/05 07:48:43

On latest motherboards integrated e1000 lan cards are PCI-E, like on my supermicro routers:

04:00.0 Ethernet controller: Intel Corp. 82573E Gigabit Ethernet Controller (Copper) (rev 03)
05:00.0 Ethernet controller: Intel Corp. 82573E Gigabit Ethernet Controller (Copper) (rev 03)

Go to intel.com, support section and download linux driver: now it is e1000-7.2.7.tar.gz

http://downloadfinder.intel.com/scripts-df-external/Detail_Desc.aspx?agr=Y&Inst=Yes&ProductID=1878&DwnldID=9180&strOSs=39&OSFullName=Linux*&lang=eng

build it: rpmbuild -ta e1000-7.2.7.tar.gz
install it: e1000-7.2.7-1.i386.rpm

reboot and integrated e1000 lan cards will work fine.

Keep in mind that intel lan cards sometimes work bad when connected to non intel/cisco switches. I resolve this issue turning off Pause frames:

ethtool -A eth0 rx off tx off

doug8429
Posts: 10
Joined: 2006/08/02 21:27:06
Location: Portland, Oregon

Re: broken e1000 network interface?

Post by doug8429 » 2006/08/09 02:52:55

Thanks. Any chance this rpm is somewhere on the net? Since the machine that needs this RPM, dosen't have an ethernet, I'm trying to build it on another machine. Also, since moving to CentOS, I stopped building kernels. I've been trying install the matching kernel source RPM, but still haven't been able to get it built. Even some clues on how to install the kernel sources would be appreciated.

Thanks in advanced.

teleport
Posts: 114
Joined: 2005/09/09 05:30:18
Contact:

Re: broken e1000 network interface?

Post by teleport » 2006/08/10 05:48:28

No problem, but rpm for which kernel? You say: "but with both the stock kernel from the 4.3 install, and now with a vanilla kernel (2.6.16.20)". Stock kernel for 4.3 are 2.6.9-34 and have two updates: 2.6.9-34.0.1 and 2.6.9-34.0.2. If You keep vanilla kernel, then module must be build against its sources.

To build kernel module is needed kernel-devel package and its dependencies.

doug8429
Posts: 10
Joined: 2006/08/02 21:27:06
Location: Portland, Oregon

Re: broken e1000 network interface?

Post by doug8429 » 2006/08/13 21:42:40

Again, thanks for the info. However, I've got (what may be a stupid) question...

If I build/install this driver for the vanilla kernel (2.6.16.20), what happens if I update (via yum) a new kernel? Will I need to rebuild/install the driver again for the new kernel? Also, I understand that this driver is somewhat new. What is the process to get it included with future kernel updates?

Thanks again.

wazzgod
Posts: 1
Joined: 2009/09/20 04:26:55
Contact:

Re: broken e1000 network interface?

Post by wazzgod » 2009/09/20 04:34:17

I am pulling my hair out for about 12 hours now.

i have a win2k8 server with hyper-v

Installed my cent os 4.7 server cd. and tryed to get eth0 going.

that was a joke.

Downloaded 5.3 dvd installed, and failed.

downloaded rpm-build and installed it (after makeing a iso and mounting)
note : cant connect to internet! on vm!

downloaded intels drivers for e1000 ect.
made iso mount again

rpmbuild and rpm -i ran
modprobe.conf
alias eth0 e1000

rebooted.
Failed.

modprobe e1000
Failed!

did same with e1000e
Failed!

Do anyone have any experience with hyper-v

Because i have done extensive research on this subject and well.
It seems hyper-v uses hosts adapter external network.

Thus gives VM's direct access to internet.

So according to everything i read.
The linux machine should be useing the e1000 driver.

CENT OS 5.3
2.6.18-128.el5 Kernel

Anyone have any ideas?
Im going to lose my mind!

Also, this is server (old fashion console ftw)
No GUI!

Zatupitel
Posts: 3
Joined: 2008/09/11 21:43:59

Re: broken e1000 network interface?

Post by Zatupitel » 2009/09/30 15:26:51

[quote]
wazzgod wrote:

Downloaded 5.3 dvd installed, and failed.

downloaded rpm-build and installed it (after makeing a iso and mounting)
note : cant connect to internet! on vm!

downloaded intels drivers for e1000 ect.
made iso mount again

rpmbuild and rpm -i ran
modprobe.conf
alias eth0 e1000

rebooted.
Failed.

modprobe e1000
Failed!

did same with e1000e
Failed!

Do anyone have any experience with hyper-v

Because i have done extensive research on this subject and well.
It seems hyper-v uses hosts adapter external network.

Thus gives VM's direct access to internet.

So according to everything i read.
The linux machine should be useing the e1000 driver.

CENT OS 5.3
2.6.18-128.el5 Kernel

Anyone have any ideas?
Im going to lose my mind!

Also, this is server (old fashion console ftw)
No GUI![/quote]

Hmm....
CENT OS 5.3 have driver e1000e (for PCI-E) :
filename: e1000e.ko
version: 1.0.2-k2
license: GPL
description: Intel(R) PRO/1000 Network Driver
But not support NAPI (kernel < 2.6.20)

This driver work!
After update kernel: 2.6.18-164.el5 #1 SMP Thu Sep 3 03:28:30 EDT 2009 x86_64 x86_64 x86_64 GNU/Linux - WORK!

eth1 Link encap:Ethernet HWaddr 00:15:17:90:37:63
inet6 addr: fe80::215:17ff:fe90:3763/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:569967292 errors:0 dropped:2585 overruns:0 frame:0
TX packets:520229642 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:434682465003 (404.8 GiB) TX bytes:330536873027 (307.8 GiB)
Memory:fde80000-fdea0000

Mannyg
Posts: 1
Joined: 2009/11/12 21:00:49

Re: broken e1000 network interface?

Post by Mannyg » 2009/11/12 21:12:18

Mine is working - First you have to create a legacy network adapter under Hyper-V management -
Start CentOs - check it connects to internal servers and then also it might be that it can connect externally.
Then, turn off centOS VM and go to the settings again in Hyper-V Management and add a Network Adapter - DO NOT DELETE THE LEGACY NETWORK ADAPTER.
Start the CentOS VM. Walaaa! You got network connection both internally and externally.

I believe the reason for this is that if you create a Network Adapter and start CentOs it will not be recognize because of the virtual driver. However, once you create a legacy adapter and start the machine it will immediately see it and configure it properly. Then the configuration is store so when you add the network adapter it will be automatically chosen out in the configuration - it will use it as the default.


Hope this help.! :)
:-D

Post Reply

Return to “CentOS 4 - Hardware Support”