We are going to add some IP addresses to make things easy to reference.

We would likely use DNS to help handle failures at the colocation site/ datacenter level. If that 2U Twin chassis is taken offline, HAProxy will detect that VM1 is offline and the second pair of servers and VM2 will take the load. These could be, for example nodes in a Supermicro 2U Twin style machine. If one server goes down, the second server can continue serving that VM. The key architecture point here is that our test Proxmox VE 4.0 4-node cluster with Ceph storage can easily be split into housing the VMs on separate pairs of servers. Here is a diagram of what we are going to do: pfSense HTTP HAProxy – game plan overview Instead of doing 1:1 NAT to a HAProxy VM, we are going to run HAProxy on pfSense which makes everything simple. We have a /30 routing a block of IPs so the WAN IP address and our publicly routable IPs are on different subnets. There are a few steps we need to take in order to set this up. It is possible to run a FreeBSD jail with HAProxy or run it on another VM, however we are going to use the pfSense VM to serve the traffic.

We are using the latest pfSense 2.2.5 release with the HAProxy-1_5 package. In this article, we will be primarily concerned with simple HTTP traffic but HTTPS offload is on our list of to-dos as well. The pfSense SG-4680 has QuickAssist so it can handle decent SSL offload as well. We are using the pfSense SG-4860 1U which has an Intel Atom C2558 onboard in one of our datacenter labs. There are many guides out there but they tend to be from older package versions. As a response to a forum member request, we are going to show how one can turn two virtual machines into a load balanced HA set. This guide will show you how to use the pfSense HAProxy package to get HA working with your web server. PfSense HTTP HAProxy - game plan overview

Pfsense haproxy