Centos 6.2 eth0 stops responding, ADDRCONF(NETDEV_UP): eth0: link is not ready.

This is becoming a Saga.  One of my new server went offline 2 days ago.  I logged in through a serial session and everything was fine apart from the networking not working.

/etc/rc.d/init.d/network restart

Failed…  Telling the cable wasn’t connected.   It was and the link was live.

A hard reboot brought it back up and I kept my fingers crossed it was a one off.  I knew at the back of my mind it wasn’t but I kept my fingers crossed anyway.

After much searching of the net I found this is a known problem across a lot of Redhat based distributions and affects the e1000e driver.

Last night the server went down again.  I found that the EPEL repo has an allegedly fixed driver.  I installed this and watched it for 2 hours.

rpm --import http://elrepo.org/RPM-GPG-KEY-elrepo.org
rpm -Uvh http://elrepo.org/elrepo-release-6-4.el6.elrepo.noarch.rpm
yum update
yum install kmod-e1000e
/sbin/shutdown -r now

I then went to bed at around 2AM this morning.  I was woken by the klaxon alarm on my BB telling me a server was down.  Yup the fix didn’t fix it.

Further research led me to add this line to my kernel options line in /boot/grub/grub.conf

pcie_aspm=off

Update: This worked.

2 Comments

Leave a Reply

Your email address will not be published.


*


This site uses Akismet to reduce spam. Learn how your comment data is processed.