Eventually we may decide to expand our cluster of PostgreSQL servers to accommodate more traffic, further increase availability, or retire an old system. Once we've established an etcd + HAProxy + Patroni stack, how difficult is the process of adding further nodes?
We wish it were possible to follow the previous recipes and consider ourselves finished. Unfortunately, modifying an operating cluster stack requires a small amount of finesse. Luckily, the extra steps are somewhat minimal, and our reward is an adaptable architecture.
Let's get started.
This recipe is somewhat unique. It depends primarily on the Installing and configuring etcd, Installing and configuring Patroni, and Installing and configuring HAProxy recipes. However, we must stress that they should not be followed exactly. The steps outlined here will explain necessary deviations, so pay close attention.