Hello Alan,
The main reason for which I suspect some sort of incompatibility or
kernel issue is due to the fact that on the desktop on which I installed
the second e1g44etblk originally run ubuntu 10.04.
There the e1g44etblk card worked when I tested it: I cannot tell which
kernel or igb version since I removed the HDD and put another one where
I installed SL 6.0: if the card was working on a different distro I
reasoned, there should something wrong with SL 6.0 .
I'll double check on monday the kernel version and igb version of the
ubuntu HDD. Now it is possible that I made a mistake and that iperf was
using eth0 and not eth1 on ubuntu, but I'm almost certain I checked
correctly (i.e. I did a ifconfig eth0 down before testing iperf): I'll
unplug eth0 as well just to be sure.
As well I got two e1g44etblk having the same problem, so I thought
that 2 over 2 with hardware problems would be really unfortunate: I use
cat6 certified cables, and i checked back to back server connection and
nic to switch connection for the iperf tests...
About my request about the kernel, well, i checked sl-testing and
fastbug and I did not see any new kernel, I was just wondering if there
was some other repo that could be used with SL. I come from CentOS and I
am aware of the problems that come with mixing repos, so I wanted to see
if there was another recommended source for SL rpms.
I'm sorry if it looked as a pressing request, my intent was not in that
direction, you have been very kind in extending an helping hand.
I'll eventually try as well to compile the last kernel from sources on
monday.. long time I did not do that, from my gentoo past days.. ah, the
memories ;)
Thanks again and have a great WE.
Cheers,
Raimondo
On 01/04/2011 23:56, Alan Bartlett wrote:
On 1 April 2011 21:34, Raimondo Giammanco [VKICC]<giamm...@vki.ac.be> wrote:
The other box where i tried the second e1g44etblk is a normal desktop with
a single pci-e slot, the motherboard is a common DQ965GF.
I've installed the el-repo repository: the version of the driver is 2.4.13,
the same of the one I've installed with the intel tar.gz and rpmbuild on the
supermicro.
I've installed the driver as for your suggestion, rebooted the machine and
tried iperf with this new driver:
before the iperf:
#############
lspci | grep -i ether
00:19.0 Ethernet controller: Intel Corporation 82566DM Gigabit Network
Connection (rev 02)
03:00.0 Ethernet controller: Intel Corporation 82576 Gigabit Network
Connection (rev 01)
03:00.1 Ethernet controller: Intel Corporation 82576 Gigabit Network
Connection (rev 01)
04:00.0 Ethernet controller: Intel Corporation 82576 Gigabit Network
Connection (rev 01)
04:00.1 Ethernet controller: Intel Corporation 82576 Gigabit Network
Connection (rev 01)
############
############
ethtool -i eth1
driver: igb
version: 2.4.13
firmware-version: 1.2-1
bus-info: 0000:03:00.0
###########
###########
rpm -qa | grep kmod
kmod-igb-2.4.13-1.el6.elrepo.x86_64
###########
after the iperf:
###########
From the machine from which I was doing the iperf -c:
when doing on eth0 (the onboard nic iwth e1000 driver)
[ 4] local 10.1.0.134 port 5001 connected with 10.1.0.131 port 49729
[ 4] 0.0-10.0 sec 1.10 GBytes 942 Mbits/sec
when doing on eth1 (the e1g44etblk)
[ 5] local 10.1.0.134 port 5001 connected with 10.1.0.134 port 51211
[ 5] 0.0-259.1 sec 86.1 MBytes 2.79 Mbits/sec
again a lot of errors in ifconfig:
eth1 Link encap:Ethernet HWaddr 00:1B:21:7C:4E:0C
inet addr:10.1.0.134 Bcast:10.1.7.255 Mask:255.255.248.0
UP BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:66919 errors:1322849926860 dropped:330712481715
overruns:330712797030 frame:1322849926860
TX packets:16073 errors:661424963430 dropped:0 overruns:0
carrier:661424963430
collisions:330712481715 txqueuelen:1000
RX bytes:94263829 (89.8 MiB) TX bytes:1061138 (1.0 MiB)
lspci changes as well:
00:19.0 Ethernet controller: Intel Corporation 82566DM Gigabit Network
Connection (rev 02)
03:00.0 Ethernet controller: Intel Corporation 82576 Gigabit Network
Connection (rev ff)
03:00.1 Ethernet controller: Intel Corporation 82576 Gigabit Network
Connection (rev ff)
04:00.0 Ethernet controller: Intel Corporation 82576 Gigabit Network
Connection (rev ff)
04:00.1 Ethernet controller: Intel Corporation 82576 Gigabit Network
Connection (rev ff)
even if ethtool -i seems correct:
###############
ethtool -i eth1
driver: igb
version: 2.4.13
firmware-version: 1.2-1
bus-info: 0000:03:00.0
###############
Could it be related to the kernel version? In the standard repos there are
no updated kernels, at least that I can see. Can you suggest any newer
kernel version and where to find it?
Raimondo,
I am beginning to suspect that it is a hardware rather than a software
issue. You have obviously noticed the apparent change of the revision
of the controller -- from 01 (before invoking iperf) to ff (after
invocation). I would not expect there to be any change whatsoever in
that field.
Would it be possible to swap out that controller for another?
As for kernels, you will find that there are the following available
from the SL repo --
[quote]
kernel-2.6.32-71.el6
kernel-2.6.32-71.7.1.el6
kernel-2.6.32-71.14.1.el6
kernel-2.6.32-71.18.1.el6
kernel-2.6.32-71.18.2.el6
[/quote]
I have the changelog deltas for each of those kernels available
(http://www.centos.toracat.org/ajb/kernel-clog-diff/el6/) but from a
quick look, I do not see anything relevant.
It's interesting that you ask me about a newer kernel version, for
those who know me are aware that I usually have a newer kernel or two
available -- but for EL5, not EL6. In all honesty, I do not think this
is either a kernel or a driver problem. As I said above, I suspect the
card itself.
Sorry that I can not "wave my wand" and resolve this for you. :-/
Alan.
