PostgreSQL 12 High Availability Cookbook - Third Edition

How to do it...

This recipe will assume that we have two PostgreSQL BDR servers as pgha1 and pgha2 and a PgBouncer and HAProxy server as pgha-proxy. Follow these steps to observe and improve cluster failover time:

1. Start the following command on pgha-proxy in a terminal window:

while [ true ]; do \     psql -qAt -U postgres myapp \         -c "SELECT inet_server_addr(), now()"; \ done

2. Open a terminal connection to both pgha1 and pgha2 with the rights to start and stop the PostgreSQL service.

3. Observe the IP address returned while step 1 operates, and stop PostgreSQL on that server. Debian-related systems can use the following command:

sudo systemctl stop postgresql@10-main

4. Red Hat systems will need to stop PostgreSQL in the following way:

sudo ...