Christopher Hunt wrote: > Ken, > I think Pasi's on to something there, I bet the GATEWAY command > in ifcfg-eth0 is mistyped or has a syntax error. In the interem, > however, a better hack might be to move the route statement from > rc.local, which only runs at boot, to > /etc/sysconfig/network-scripts/route-eth0 . That will enable the > network service to restart or the eth0 interface to downup without > removing the default route. Thanks for this tip. I removed the GATEWAY and GATEWAYDEV from /etc/sysconfig/network which got rid of the initial (incorrect route) and added default via 192.168.144.6 dev eth0 onlink to the /etc/sysconfig/network-scripts/route-eth0 Note the use of the 'onlink' option. It appears to be working fine without the arp traffic. I guess I thought this was a more common configuration when using Xen in a routed configuration. I'm not sure if the network scripts allow a mechanism to add the 'onlink' option so that GATEWAY/GATEWAYDEV specified are actually used. Grepping through the network-scripts does not yield anything. If other people are using this in the same way I am, you will need to do this change or else there will be all the unnecessary arp cache activity for every single ip address which might fill up your arp table. _______________________________________________ CentOS-virt mailing list CentOS-virt@xxxxxxxxxx http://lists.centos.org/mailman/listinfo/centos-virt
