[SGVLUG] Routing issues?

[Rae Yip]

Are you using LVS-NAT, LVS-DR, or LVS-Tun?

-Rae.

On Tue, Mar 24, 2009 at 4:38 PM, Solomon K. Chang <skevin521 at yahoo.com> wrote:
>
> I'm having a few difficulties with the final stages of setting up an LVS cluster, and was hoping someone on this list may have done the same, as well as had the same problems as me.
>
> For those of you familiar with LVS, it's just setting up a machine that, for all intents and purposes, is just a load balancer - it receives requests (like, say, port 80), and redirects those requests to a big pool of servers (like, say, a whole bunch of webservers).  It has some additional functionality, like being able to tell when a node is down and remove that node from the pool until someone puts it back up.
>
> Only problem is, all the setup instructions, everywhere I go, tell me that the Load Balancer must have an alternative Virtual IP (hereafter known as VIP), and all the nodes must have the same address for a loopback alias.  So in essence, the load balancer has:
>
> eth0:0    10.0.0.100/255.255.255.0
>
> ...and the nodes must have:
>
> lo:0      10.0.0.100/255.255.255.255
>
> (Notice the full mask for the node loopback alias.)
> Once I set these loopback aliases on the nodes, the Load Balances can no longer get responses from these nodes.  It can *forward* a web request to the node just fine, but once the HeartBeat monitor of the load balancer detects that the nodes is down, it is removed from the list of available servers.
>
> Unfortunately, the heartbeat monitor is inextricably tied into the load balancer.  How can I get the Load Balancer to properly see the nodes?
>
>
> Thanks in advance,
> Solomon
>
>
>
>
>