Author: Dean Scothern <dean.scothern@xxxxxxxxxxxxxx>
Date: Thu, 8 Mar 2012 12:45:23 +0000
Hi, It seems that the solution to my problem involves the use of /proc/sys/net/ipv4/conf/eth<n>/accept_local For test purposes I've installed a more recent kernel that offers this functionality and I
Author: Dean Scothern <dean.scothern@xxxxxxxxxxxxxx>
Date: Fri, 2 Mar 2012 10:12:17 +0000
Thanks, It's worth a try but I think it will fall foul of the host route in the local table, which will apply first. Best Regards Had a thought on this - You could use iptables to mark your return pa
Had a thought on this - You could use iptables to mark your return packet, then run it through a separate routing table based on a 'i ip ru add fwmark x table y' option. Your separate routing table w
You can either snat or run in dr or tun mode. Sent from my iPhone _______________________________________________ Please read the documentation before posting - it's available at: http://www.linuxvir
Author: Dean Scothern <dean.scothern@xxxxxxxxxxxxxx>
Date: Thu, 1 Mar 2012 13:59:42 +0000
Thank you for your quick reply. Whilst snat would work I would prefer not to use it as it hides the source ip of the packets, making applications that use ip access lists more problem atic to configu
You need to SNAT real server traffic going to your real servers. Quick google found this: http://blog.loadbalancer.org/enabling-snat-in-lvs-xt_ipvs-and-iptables/ I'm presuming it's in mainline by now
Author: Dean Scothern <dean.scothern@xxxxxxxxxxxxxx>
Date: Thu, 1 Mar 2012 12:55:52 +0000
Hi, I've been experimenting with a slightly non standard lvs cluster arrangement. I have a set of combined real servers/real clients (each machine has both services and clients) and two machines runn