-
Greetings you great engineers,
I'm trying to get one of these intel cards working - remotely.
short intro: newly installed citrix xenserver 5.5, shipped to remote location, having serious trouble with e1000 card:
---
# lspci
00:00.0 RAM memory: nVidia Corporation MCP61 Memory Controller (rev a1)
00:01.0 ISA bridge: nVidia Corporation MCP61 LPC Bridge (rev a2)
00:01.1 SMBus...
2009-11-14 01:42:34 UTC by seggi
-
With flow control disabled, I was not able to reproduce the issue so far.
Will wait for a patch, thank you.
2009-11-12 20:25:59 UTC by Sergey Cheperis
-
please try disabling flow control using
ethtool -A eth0 rx off tx off autoneg off
and see if the issue goes away, we think we already know of this problem and have a fix. The command above works around the issue until we have an official patch.
2009-11-12 19:49:03 UTC by go_jesse
-
The machine is a file server, and when I copy a 100 MB file from this server to a workstation over NFS (transfer rate is 60-100 MBytes/sec), I get the hang in about 10% cases.
The in--kernel driver did the same.
Sure, I'll be able to run the patched driver.
I'm attaching the files with the info you requested.
2009-11-12 19:04:40 UTC by Sergey Cheperis
-
assigning dave, Sergey was there any test or traffic in particular that helps cause this hang? Did you try running the e1000e driver that came with the kernel?
we also have some low overhead debug code called e1000_dump that can dump the state of the transmit ring, would you be willing to run a patched driver?
please attach a full dmesg file from boot, as well as lspci -vvv output, and...
2009-11-12 17:13:09 UTC by go_jesse
-
[ 360.040466] 0000:03:00.0: eth0: Detected Hardware Unit Hang:
[ 360.040468] TDH
[ 360.040470] TDT
[ 360.040470] next_to_use
[ 360.040471] next_to_clean
[ 360.040472] buffer_info[next_to_clean]:
[ 360.040473] time_stamp
[ 360.040474] next_to_watch
[ 360.040475]...
2009-11-12 14:26:27 UTC by Sergey Cheperis
-
go_jesse added stephenko to the Intel Wired Ethernet project.
2009-11-05 22:22:11 UTC by go_jesse
-
Is there something we can do to confirm this? What version of the driver is 'the right one' or specifically which patches would be helpful? My understanding of the hardware is that there is no BMC and an add-on card needs to be purchased for this platform for that functionality? Is there a related bug or more documentation on the issue you are describing? Do we know why a module...
2009-11-05 00:11:31 UTC by elatt
-
I think this is likely related to the known BMC interaction issues, which may require a BMC update from your vendor, an eeprom update to the LAN eeprom, and the right driver.
2009-11-04 23:47:23 UTC by go_jesse
-
We've seen on our Supermicro X7DBT servers having issues when booting with the kernel loading the e1000e driver for all our interfaces. The issue doesn't happen all the time but it is frequent enough to cause problems in our environment for machines that get rebooted frequently. We are running Debian Lenny with a kernel from squeeze (2.6.30). Typically eth1 is the NIC that fails the probe but...
2009-11-04 23:17:00 UTC by elatt
