PostgreSQL 12 High Availability Cookbook - Third Edition

There's more...

So how would we permanently remove pgha1 from the cluster so that pgha3 replaces it entirely? We really only need to perform a handful of steps:

1. Remove pgha1 from haproxy.cfg and reload the HAProxy service.
2. Remove pgha1 from the cluster by executing the following command from either pgha2 or pgha3:

SELECT bdr.part_node('chicago'); 

3. Purge the pgha1 metadata from the cluster by running the following SQL on both pgha2 and pgha3:

SELECT bdr.drop_node('chicago'); 

4. Verify that the node has been removed with the following query:

SELECT node_seq_id, node_id, node_name, 
       node_group_name, peer_state_name 
  FROM bdr.node_summary; 

The first step should be obvious. We don't want HAProxy to send traffic to a node we're decommissioning. ...