I'm looking at the available settings here, and it's just not adding up for me. I'm hoping someone has some experience with this, or can help to fill in the gaps for me.
First, my assumptions:
1) I can use the HA cluster as an explicit proxy...I don't need to have it in-line with the default route out to the internet.
2) rather than pointing my clients to the physical IP addresses of each individual MWG appliances, I would point them to a single (shared/virtual) IP address, which would load-balance the traffic between appliances
In addition to the Common HTTP Proxy settings that I'm familiar with, the HA Proxy also has these settings:
Port Redirects - I'm not sure why this one exists, or what it's purpose is. If my clients are pointed at a (virtual) proxy address/port combination, why is this setting necessary/useful? Is it mandatory?
Director Priority - this one seems straight-forward enough. It determines the liklihood that a given proxy will take over the cluster as a director.
Management IP - Again, seems straight-forward enough. This is the IP used to monitor/communicate with the cluster nodes. Grayed out if the Director Priority is 0because that means this node will never have need to talk to it's peers.
Virtual IPs - Here's the one that's really getting me. My initial assumption was that this is the shared Cluster IP Address. I assumed that my clients could connect to this address, and the active Director node would make a determination as to which node should handle that client's requests. I'm wondering why it's asking for CIDR notation, rather than an individual IP (would a 32-bit subnet mask work here?). Is my director node supposed to be listening "in front" of the physical appliances, and should the virtual addresses correspond with the Physical IPs of the other nodes in my cluster? If not, where is it determined what nodes will participate in a given HA cluster?
Virtual Router ID - last 2 hex characters in my cluster's MAC address, presumably taken over by the active Director Node.
VRRP Interface - The interface that hosts the cluster's virtual IP(s).
Just having trouble wrapping my head around how to load balance these things, and the documentation is pretty weak with regard to this topic. Any experienced or otherwise informed input would be greatly appreciated.