PostgreSQL High Availability Cookbook - Second Edition

How it works...

We've already discussed the process of creating streaming replicas several times throughout this book, so we've elected to use a shortened version here. Our primary goal here is to create a full database clone of pg-primary on pg-primary2. This clone should continue to receive data from pg-primary until we are ready to split up our application data. When database activity is low or we can temporarily disable write activity to the myapp2 schema, we can promote pg-primary2 so that it acts as a writable server.

Once pg-primary2 is writable, we execute an UPDATE statement on the shard_map table in pg-shared. Then, we either refresh or invalidate cached copies of that table so that they are rebuilt. From this point on, all new ...