I have a problem with LVS-DR and loopback interfaces.
I have three xen servers each with a Centos 6.2 VM running apache for the real
servers. On xenserver01 I have a Centos 6.2 vm that is the primary load
balancer. On xenserver02 there is a Centos 6.2 vm that is the backup load
balancer. I am running LVS in DR. I have arp disabled on all the real servers
using the arp_announce and arp_ignore parameters in /etc/sysctl.conf.
I can do DR just fine using postrouting rules in iptables with no loopback
interfaces configured for VIPs on real servers. This works fine using name
based virtual hosts. My client requires ip based virtualhosts. In order for
apache to respond to requests on the VIP I need to configure loopback
interfaces for each of my 26 VIPs. When I configure the local loopback
interfaces the VIPs stop responding and the ipvs table shows all connections as
InActConn.
I can connect to the loopback interface and port locally from the real server.
The loadbalancer can connect to the RIP and port. I can connect to the RIPS
directly from a browser.
Is there a problem running IPVS on virtual machines?
Thanks in advance for any advice. I can provide much more detail about the
environment if needed.
PGP.sig
Description: PGP signature
_______________________________________________
Please read the documentation before posting - it's available at:
http://www.linuxvirtualserver.org/
LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
or go to http://lists.graemef.net/mailman/listinfo/lvs-users
|