In centOS there is a random bug in the stable version of the e1000e drivers.
This may cause the server to sometimes not respond, with the following showing in the kernel logs.
May 25 12:42:51 sau-cea10-or kernel: NETDEV WATCHDOG: eth0 (e1000e): transmit queue 0 timed out
In order to resolve this, you will need to utilise the below commands from the Command Line, to update to the latest version of the elRepo Driver, and install kmod-e1000e Driver, and issue a boot for best effect. (You should do this from console in most cases). This is mostly for CentOS 6, but you can use elrepo on any platform if you experience the issue.
rpm -Uvh http://www.elrepo.org/elrepo-release-6-6.el6.elrepo.noarch.rpm
yum --enablerepo=elrepo install kmod-e1000e
shutdown -r now
Once the service comes back up, it should have the updated driver, which you can test with ethtool (assuming the primary device is eth0).
ethtool -i eth0 | grep 'driver\|version'