Just installed 5.3, fresh install, but my NIC is not recognized

[wilecoyote]

Hi,

I have a new server with a P5QL-VM DO motherboard with onboard ethernet. I did a fresh install of CentOS 5.3 but my network interface was not detected. I tried reading as many relevant threads as I could find, but still haven't solved the problem - no doubt because I'm doing something dumb.

Here is some information which may help:

[root@localhost ~]# uname -a
Linux localhost.localdomain 2.6.18-92.el5 #1 SMP Tue Jun 10 18:49:47 EDT 2008 i686 i686 i386 GNU/Linux


[root@localhost ~]# /sbin/lspci | grep ther
00:19.0 Ethernet controller: Intel Corporation Unknown device 10df (rev 02)

[root@localhost ~]# /sbin/lspci -n | grep "^$(/sbin/lspci | awk '/ther/ { print $1 }')"
00:19.0 0200: 8086:10df (rev 02)

[root@localhost ~]# grep -i 8086 /lib/modules/*/modules.alias | grep -i 10df
alias pci:v00008086d000010DFsv*sd*bc*sc*i* e1000e

To get this far, I installed an rpm from elrepo called:
kmod-e1000e-1.0.2.5_NAPI_2.el5.elrepo.i686.rpm

I then tried to configure my ethernet using the ethernet configuration GUI - I wasn't totally sure how to do this but I tried adding a new device and selecting "Intel Corporation Unknown Device 10df" as the device. But after setting it up and trying to activate it, I get:

Cannot activate network device eth0!
e1000 device eth0 does not seem to be present, delaying initialization.

I would really appreciate any help. I'm pretty new with linux but willing to learn!

Peter

[AlanBartlett]

[quote]I did a fresh install of CentOS 5.3[/quote]
Yet the kernel you mention --

[quote]2.6.18-92.el5[/quote]
-- is from [i]CentOS 5.2[/i].

Taking the Vendor:Device ID pairing that you've shown (8086:10df), I see --

[code]
$ grep -i 8086 /lib/modules/*/modules.alias | grep -i 10df
/lib/modules/2.6.18-128.el5/modules.alias:alias pci:v00008086d000010DFsv*sd*bc*sc*i* e1000e
/lib/modules/2.6.18-164.el5/modules.alias:alias pci:v00008086d000010DFsv*sd*bc*sc*i* e1000e
[/code]
This tells me that the NIC requires the [i]e1000e[/i] kernel module and is only (natively) supported with kernel-2.6.18-128.el5 ([i]CentOS 5.3[/i]) and upwards.

I now go and check the [url=http://elrepo.org/tiki/DeviceIDs]DeviceIDs page[/url] at [url=//elrepo.org]ELRepo[/url] and find the following line referencing that Vendor:Device ID pairing --

[code]
pci 8086:10DF kmod-e1000e
[/code]
This tells me that a download and installation of the [i]kmod-e1000e[/i] package -- currently [i]kmod-e1000e-1.0.2.5_NAPI_2.el5.elrepo.i686.rpm[/i] -- is appropriate for your 32-bit system.

So far, everything you've done looks good, [b]Peter[/b].

And now we've got to the [i]phoo-bah[/i]. Please take a look in your [b]/etc/modprobe.conf[/b] file. You should have a line that reads --

[code]
[b]alias eth0 e1000e[/b]
[/code]
I suspect, reading the error message that you've received --

[quote]
Cannot activate network device eth0!
e1000 device eth0 does not seem to be present, delaying initialization.
[/quote]
-- that the trailing [b]e[/b] is missing from the [i]alias[/i] line and the [i]eth0[/i] device is, as a consequence, being associated with the [i]e1000[/i] module. :roll:

So, correct the line that references the [i]eth0[/i] device in your [b]/etc/modprobe.conf[/b] file and then restart the network services.

As [i]root[/i] --

[code]
[b]service network restart[/b]
[/code]
If you are still having problems, please post the output returned by the following --

[code]
[b]rpm -q centos-release
rpm -qa kmod\*
grep eth0 /etc/modprobe.conf
ls -l $(find /lib/modules -name \*.ko | grep -E 'extra|weak')[/b]
[/code]
Once you have working access to the Internet, your first task will be to read the [url=http://wiki.centos.org/Manuals/ReleaseNotes/CentOS5.3]CentOS 5.3 release notes[/url] and then update your system. :-)

[wilecoyote]

Hi Alan,

First of all, thank you very much for your help!




[quote]
AlanJBartlett wrote:
[quote]I did a fresh install of CentOS 5.3[/quote]
Yet the kernel you mention --

[quote]2.6.18-92.el5[/quote]
-- is from [i]CentOS 5.2[/i].
[/quote]

I'm surprised because the name of the iso file I used was CentOS-5.3-i386-bin-DVD.iso. Am I missing something?

[quote]
I suspect, reading the error message that you've received --

[quote]
Cannot activate network device eth0!
e1000 device eth0 does not seem to be present, delaying initialization.
[/quote]
-- that the trailing [b]e[/b] is missing from the [i]alias[/i] line and the [i]eth0[/i] device is, as a consequence, being associated with the [i]e1000[/i] module. :roll:

So, correct the line that references the [i]eth0[/i] device in your [b]/etc/modprobe.conf[/b] file and then restart the network services.
[/quote]

You were correct - the [b]/etc/modprobe.conf[/b] line was missing the trailing "e" (is that something I caused?). I fixed the line, and restarted the network service, but still got this error:

[code]
Bringing up interface eth0: e1000e device eth0 does not seem to be present, delaying initialization.
[FAILED]
[/code]

So here is the output to the commands you requested:

[code]
[root@localhost ~]# rpm -q centos-release
centos-release-5-2.el5.centos

[root@localhost ~]# rpm -qa kmod\*
kmod-e1000e-1.0.2.5_NAPI-2.el5.elrepo

[root@localhost ~]# grep eth0 /etc/modprobe.conf
alias eth0 e1000e

[root@localhost ~]# ls -l $(find /lib/modules -name \*.ko | grep -E 'extra|weak')
-rw-r--r-- 1 root root 157016 Oct 9 09:34 /lib/modules/2.6.18-8.el5/extra/e1000e/e1000e.ko
lrwxrwxrwx 1 root root 48 Oct 16 18:21 /lib/modules/2.6.18-92.el5/weak-updates/e1000e/e1000e.ko -> /lib/modules/2.6.18-8.el5/extra/e1000e/e1000e.ko

[/code]

Again, thanks so much for your help. If this is a pain, I could always start again from scratch with the latest (5.3) install, if I can figure out why the one I downloaded was 5.2!

Peter

[toracat]

If you have not done so, try rebooting. Then inspect dmesg and see if the system identified the NIC.

[wilecoyote]

[quote]
toracat wrote:
If you have not done so, try rebooting. Then inspect dmesg and see if the system identified the NIC.[/quote]

Thanks, but no luck. It still fails. Here is what I found with dmesg after the reboot:

[code]
[root@localhost ~]# dmesg | grep e1000
e1000e: Intel(R) PRO/1000 Network Driver - 1.0.2.5-NAPI
e1000e: Copyright(c) 1999 - 2009 Intel Corporation.
e1000e: probe of 0000:00:19.0 failed with error -5

[/code]

[wilecoyote]

Update: Success!

I downloaded a different iso image of the 5.3 DVD, and this time it actually [b]was[/b] the 5.3 release and not the 5.2 release. I then tried doing an upgrade over top of the 5.2 release, but I ended up with the same problem.

So, since I had nothing to save (this was a fresh install on new hardware), I tried a completely fresh install with the 5.3 release, and that worked perfectly.

In case it helps anyone else, here are the results from the various commands that you had me try, now with the working system:
[code]
[root@server3 ~]# uname -a
Linux server3.3dbresearch.com 2.6.18-128.el5 #1 SMP Wed Jan 21 10:44:23 EST 2009 i686 i686 i386 GNU/Linux

[root@server3 ~]# rpm -qa kernel\* | sort
kernel-2.6.18-128.el5

[root@server3 ~]# /sbin/lspci | grep ther
00:19.0 Ethernet controller: Intel Corporation 82567LF-3 Gigabit Network Connection (rev 02)

[root@server3 ~]# /sbin/lspci -n | grep "^$(/sbin/lspci | awk '/ther/ { print $1 }')"
00:19.0 0200: 8086:10df (rev 02)

[root@server3 ~]# rpm -qa kernel\* | sort
kernel-2.6.18-128.el5

[root@server3 ~]# /sbin/lspci -n | grep "^$(/sbin/lspci | awk '/ther/ { print $1 }')"
00:19.0 0200: 8086:10df (rev 02)

[root@server3 ~]# rpm -q centos-release
centos-release-5-3.el5.centos.1

[root@server3 ~]# grep eth0 /etc/modprobe.conf
alias eth0 e1000e

[root@server3 ~]# ls -l $(find /lib/modules -name \*.ko | grep -E 'extra|weak')
total 68
-rw------- 1 root root 1399 Oct 17 12:42 anaconda-ks.cfg
drwxr-xr-x 2 root root 4096 Oct 17 13:03 Desktop
-rw-r--r-- 1 root root 32693 Oct 17 12:42 install.log
-rw-r--r-- 1 root root 5385 Oct 17 12:42 install.log.syslog
-rw-r--r-- 1 root root 195 Oct 17 12:59 scsrun.log
[/code]


I really appreciate everyone's help, and if there's anything else you think I should post to help others, just let me know. And if it's obvious to anyone why my attempt to install the driver under 5.2 failed, I'd love to know.


Peter

[AlanBartlett]

I'm pleased to read of your success, [b]Peter[/b].

[quote]if it's obvious to anyone why my attempt to install the driver under 5.2 failed, I'd love to know.[/quote]
It's not at all obvious and without the exact hardware to experiment with, I am unable to reproduce that negative result. :-(

[quote]e1000e: probe of 0000:00:19.0 failed with error -5[/quote]
That is puzzling. I'm now wondering what the kernel message "[i]error -5[/i]" signifies . . .

Although you have only just installed the system, please remember to now execute a [b]yum update[/b] command to bring it fully up to date. ;-)

[wilecoyote]

[quote]
[quote]e1000e: probe of 0000:00:19.0 failed with error -5[/quote]
That is puzzling. I'm now wondering what the kernel message "[i]error -5[/i]" signifies . . .
[/quote]

I guess I just assumed that there were other things in the 5.3 release that were required to make this NIC work properly, not just the drivers. Also,, it's completely puzzling to me why my first download of what seemed to be the 5.3 release was actually the 5.2 release. Maybe there were other issues with that first iso file I downloaded?

I do have a 2nd system with this identical hardware. If there was something specific you wanted me to try that would help out in any way, just let me know.

[quote]
Although you have only just installed the system, please remember to now execute a [b]yum update[/b] command to bring it fully up to date. ;-)
[/quote]

I've done this and it all seemed to work well. Thanks again for all your help! This was actually just a dry run for me to make sure I could get 5.3 up and running on this new hardware. Now I'm about to build our real server, so I'm busy trying to make decisions about the partitioning scheme, whether to use Linux RAID 1, and whether to use LVM. Lots to learn!

Thanks again,
Peter

[AlanBartlett]

Just for the record, there was a bug in the e1000e driver (version 1.0.2.5).

[i]Intel[/i] have updated it to version 1.0.15 and the [i]ELRepo[/i] kmod packages have been rebuilt. :-)