Introduction

High availability (HA) is a critical requirement for modern database infrastructures, ensuring minimal downtime and seamless failover in case of node failures. PostgreSQL, while robust as a standalone database, requires additional tooling to manage replication, failover, and cluster orchestration in multi‑node environments.

repmgr (Replication Manager) is a widely adopted open‑source tool that simplifies the administration of PostgreSQL replication and failover. Version 5.5 of repmgr introduces enhanced support for PostgreSQL 16.x, making it a reliable choice for production‑grade HA setups.

This guide focuses on deploying PostgreSQL 16.11 with repmgr 5.5 on CentOS Stream 10 (Coughlan), the rolling‑release distribution aligned with Red Hat Enterprise Linux 10. The target architecture is a three‑node cluster consisting of one primary and two standby nodes, configured for synchronous replication and automatic failover

Cluster Topology: Three nodes

• Node 1 (Primary): Handles read/write traffic and replicates changes downstream.
• Node 2 (Standby): Hot standby, ready to be promoted in case of primary failure.
• Node 3 (Witness): Provides quorum and ensures consistent failover decisions.

On a host:

Checking OS Version:

Command:

Output:

NAME="CentOS Stream" VERSION="10 (Coughlan)" ………………… ……………… 1 2 3 4 5 6 7 NAME="CentOS Stream"   VERSION="10 (Coughlan)"   …………………   ………………

Installing Prerequisites for the repmgr:

Untar repmgr package:

tar -xvf repmgr-5.5.0.tar.gz cd repmgr-5.5.0 which pg_config 1 2 3 tar -xvf repmgr-5.5.0.tar.gz cd repmgr-5.5.0 which pg_config

Output:

Installing repmgr in pg_config path:

export PG_CONFIG=/usr/pgsql-16/bin/pg_config ./configure make sudo make install 1 2 3 4 export PG_CONFIG=/usr/pgsql-16/bin/pg_config ./configure make sudo make install Added in bash_profile export PATH=/usr/pgsql-16/bin:$PATH echo 'export PATH=/usr/pgsql-16/bin:$PATH' >> ~/.bash_profile source ~/.bash_profile 1 2 3 export PATH=/usr/pgsql-16/bin:$PATH echo 'export PATH=/usr/pgsql-16/bin:$PATH' >> ~/.bash_profile source ~/.bash_profile

 

[postgres@test3 data]$ repmgr --version repmgr 5.5.0 [postgres@test3 data]$ repmgrd --version repmgrd 5.5.0 1 2 3 4 5 [postgres@test3 data]$ repmgr --version repmgr 5.5.0   [postgres@test3 data]$ repmgrd --version repmgrd 5.5.0

Prerequisites:

Enable ssh in all the three machine:

ssh-keygen -t ed25519 ssh-copy-id postgres@192.168.134.129 ssh-copy-id postgres@192.168.134.130 ssh postgres@192.168.134.130 "hostname" ssh postgres@192.168.134.131 "hostname" 1 2 3 4 5 6 7 ssh-keygen -t ed25519   ssh-copy-id postgres@192.168.134.129 ssh-copy-id postgres@192.168.134.130   ssh postgres@192.168.134.130 "hostname" ssh postgres@192.168.134.131 "hostname"

 

Added entry for allowing port 5432 in tcp

Verify:

Adding host in all the three node in /etc/hosts :

Repmgr Setup:

Creating user in the primary and witness node:

CREATE USER repmgr WITH REPLICATION SUPERUSER LOGIN; ALTER USER repmgr SET search_path TO repmgr, public; CREATE DATABASE repmgr OWNER repmgr; 1 2 3 CREATE USER repmgr WITH REPLICATION SUPERUSER LOGIN; ALTER USER repmgr SET search_path TO repmgr, public; CREATE DATABASE repmgr OWNER repmgr;

In postgresql.conf add following configuration changes in primary (node1) and (node2) standby :

listen_addresses = '*' port = 5432 max_wal_senders = 10 max_replication_slots = 10 wal_level = replica hot_standby = on archive_mode = on archive_command = '/bin/true' # or your preferred archiving command wal_log_hints = on shared_preload_libraries = 'repmgr' 1 2 3 4 5 6 7 8 9 10 listen_addresses = '*' port = 5432 max_wal_senders = 10 max_replication_slots = 10 wal_level = replica hot_standby = on archive_mode = on archive_command = '/bin/true' # or your preferred archiving command wal_log_hints = on shared_preload_libraries = 'repmgr'

 

Added entry for the repmgr in pg_hba.conf

In a primary node create /etc/repmgr.conf

Node 01 (primary)

node_id=1 node_name='node1' conninfo='host=192.168.134.129 user=repmgr dbname=repmgr connect_timeout=2' data_directory='/var/lib/pgsql/16/data' pg_bindir='/usr/pgsql-16/bin' config_directory='/var/lib/pgsql/16/data/postgresql.conf' replication_user='repmgr' failover=automatic promote_command='repmgr standby promote -f /etc/repmgr.conf --log-to-file' follow_command='repmgr standby follow -f /etc/repmgr.conf --log-to-file --upstream-node-id=%n' log_file='/var/log/repmgr/repmgr.log' 1 2 3 4 5 6 7 8 9 10 11 node_id=1 node_name='node1' conninfo='host=192.168.134.129 user=repmgr dbname=repmgr connect_timeout=2' data_directory='/var/lib/pgsql/16/data' pg_bindir='/usr/pgsql-16/bin' config_directory='/var/lib/pgsql/16/data/postgresql.conf' replication_user='repmgr' failover=automatic promote_command='repmgr standby promote -f /etc/repmgr.conf --log-to-file' follow_command='repmgr standby follow -f /etc/repmgr.conf --log-to-file --upstream-node-id=%n' log_file='/var/log/repmgr/repmgr.log'

 

Node 02 (secondary)

sudo vi /etc/repmgr.conf

node_id=2 node_name='node2' conninfo='host=192.168.134.130 user=repmgr dbname=repmgr connect_timeout=2' data_directory='/var/lib/pgsql/16/data' pg_bindir='/usr/pgsql-16/bin' config_directory='/var/lib/pgsql/16/data/postgresql.conf' replication_user='repmgr' failover=automatic promote_command='repmgr standby promote -f /etc/repmgr.conf --log-to-file' follow_command='repmgr standby follow -f /etc/repmgr.conf --log-to-file --upstream-node-id=%n' log_file='/var/log/repmgr/repmgr.log' 1 2 3 4 5 6 7 8 9 10 11 node_id=2 node_name='node2' conninfo='host=192.168.134.130 user=repmgr dbname=repmgr connect_timeout=2' data_directory='/var/lib/pgsql/16/data' pg_bindir='/usr/pgsql-16/bin' config_directory='/var/lib/pgsql/16/data/postgresql.conf' replication_user='repmgr' failover=automatic promote_command='repmgr standby promote -f /etc/repmgr.conf --log-to-file' follow_command='repmgr standby follow -f /etc/repmgr.conf --log-to-file --upstream-node-id=%n' log_file='/var/log/repmgr/repmgr.log'

Node 03 (Witness)

Sudo vi /etc/repmgr.conf

node_id=3 node_name='witness' conninfo='host=192.168.134.131 user=repmgr dbname=repmgr connect_timeout=2' data_directory='/var/lib/pgsql/16/data' pg_bindir='/usr/pgsql-16/bin' replication_user='repmgr' log_file='/var/log/repmgr/repmgr.log' 1 2 3 4 5 6 7 node_id=3 node_name='witness' conninfo='host=192.168.134.131 user=repmgr dbname=repmgr connect_timeout=2' data_directory='/var/lib/pgsql/16/data' pg_bindir='/usr/pgsql-16/bin' replication_user='repmgr' log_file='/var/log/repmgr/repmgr.log'

Create Log Directory

In all nodes create log directory

sudo mkdir -p /var/log/repmgr sudo chown postgres:postgres /var/log/repmgr 1 2 sudo mkdir -p /var/log/repmgr sudo chown postgres:postgres /var/log/repmgr

Register nodes in a cluster:

Register Primary Node in a repmgr cluster:

Building standby from the primary:

Register standby node:

Register Witness node in a repmgr cluster make sure datadirectory name is different from primary:

Register repmgrd service in all the three node:

[Unit] Description=A replication manager, and failover management tool for PostgreSQL After=syslog.target After=network.target After=postgresql.service [Service] Type=forking User=postgres ExecStart=/usr/pgsql-16/bin/repmgrd -f /etc/repmgr.conf --pid-file=/var/run/repmgrd.pid --daemonize ExecReload=/bin/kill -HUP $MAINPID PIDFile=/var/run/repmgrd.pid [Install] WantedBy=multi-user.target 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 [Unit] Description=A replication manager, and failover management tool for PostgreSQL After=syslog.target After=network.target After=postgresql.service   [Service] Type=forking User=postgres ExecStart=/usr/pgsql-16/bin/repmgrd -f /etc/repmgr.conf --pid-file=/var/run/repmgrd.pid --daemonize ExecReload=/bin/kill -HUP $MAINPID PIDFile=/var/run/repmgrd.pid   [Install] WantedBy=multi-user.target

Reload repmgrd and its services:

Verify repmgr running and checking status:

Conclusion:

Deploying a PostgreSQL 16.11 high-availability cluster using repmgr 5.5 on CentOS Stream 10 showcases the ability of open-source tools to provide enterprise-level reliability. The setup involves a primary node, a hot standby, and a witness server to enable synchronous replication and automatic failover. The detailed setup process emphasizes the significance of planning and reproducible configuration. The cluster supports seamless failover and minimal downtime, making it suitable for critical workloads. With improved support for PostgreSQL 16.x, repmgr offers an effective solution for managing replication and failover, creating a reliable and scalable HA platform for modern data infrastructures.