Step-by-Step Guide to PostgreSQL Streaming Replication using RepMgr

Streaming replication in PostgreSQL ensures high availability and scalability by maintaining multiple copies of a database and continuously synchronizing standby nodes with the primary node. This guide walks you through setting up asynchronous streaming replication using RepMgr, making it applicable to various PostgreSQL versions.

Overview of the Setup

The objective is to configure asynchronous streaming replication between a primary node and multiple standby nodes to enhance efficiency and reliability. This setup ensures that changes are copied from the primary node to standby nodes without waiting for confirmation, improving performance.

Fig: Streaming Replication in PostgreSQL

Replication: What is it?

Replication in PostgreSQL is the process of maintaining synchronized copies of a database. It is essential for high availability, load balancing, and disaster recovery. PostgreSQL supports two primary replication methods:

• Physical (Streaming) Replication: Real-time data replication from the primary to standby databases. The primary handles both read and write operations, while standbys serve as read-only replicas.

• Logical Replication: Uses a publish-subscribe model, allowing selective replication of tables, rows, or columns across different PostgreSQL versions.

In this guide, we will focus on asynchronous streaming replication for high performance.

Step 1: Install PostgreSQL on All Standby Servers

Before setting up replication, install PostgreSQL on all the standby nodes.

a. Add the PostgreSQL Repository

sudo sh -c 'echo "deb http://apt.postgresql.org/pub/repos/apt/ $(lsb_release -cs)-pgdg main" >> /etc/apt/sources.list.d/pgdg.list'
wget -q https://www.postgresql.org/media/keys/ACCC4CF8.asc -O - | sudo apt-key add -

b. Install PostgreSQL

sudo apt-get update
sudo apt install postgresql

c. Verify Installation

Ensure that the standby nodes have the same version of PostgreSQL as the primary node to maintain compatibility and consistency.

psql --version

Step 2: Create a Data Directory for the Cluster

Create a dedicated directory on all standby servers for the PostgreSQL cluster.

sudo mkdir -p /<DATA_PATH>/<CLUSTER_NAME>

Note: Replace <DATA_PATH> with the base directory and <CLUSTER_NAME> with a name for the cluster.

Step 3: Initialize the PostgreSQL Cluster on the Standby Node

Initialize a new PostgreSQL cluster on the standby node.

sudo pg_createcluster <VERSION> -p <CUSTOM_PORT> -d <DATA_PATH> <CLUSTER_NAME>

Note: Replace <CUSTOM_PORT> with the desired port number and replace<VERSION> with the PostgreSQL version.

Step 4: Edit PostgreSQL Configuration

Modify postgresql.conf on both primary and standby nodes.

File location:

/etc/postgresql/<VERSION>/<CLUSTER_NAME>/postgresql.conf

Adjust key parameters:

listen_addresses = '*'
port = <CUSTOM_PORT>
wal_level = 'hot_standby'

Note: If the PostgreSQL version is 9 or lower, set wal_level to ‘hot_standby’. For versions 10 and above, set wal_level to ‘replica’.

Customize additional settings (e.g., shared_buffers, work_mem) based on your workload.

Step 5: Configure pg_hba.conf for Replication Access on the standby servers

Modify pg_hba.conf to allow trusted connections for replication.

File location:

/etc/postgresql/<VERSION>/<CLUSTER_NAME>/pg_hba.conf

Add entries for primary and standby nodes:

host replication <REPLICATION_USER> <PRIMARY_NODE_IP>/32 md5
host replication <REPLICATION_USER> <STANDBY_NODE_IP>/32 md5

Step 6: Create a Replication User on the Primary Node

Create a dedicated replication user.

sudo -u postgres createuser -s replicator -p <CUSTOM_PORT>
sudo -u postgres psql -p <CUSTOM_PORT> -c "ALTER USER replicator WITH PASSWORD '<YOUR_PASSWORD>';"

Step 7: Install RepMgr on All Nodes

a. Add the RepMgr Repository

curl https://dl.2ndquadrant.com/default/release/get/deb | sudo bash

b. Install RepMgr

sudo apt install postgresql-repmgr

c. Verify Installation

/usr/lib/postgresql/<VERSION>/bin/repmgr --version

Step 8: Create the RepMgr User and Database on the Primary Node

Set up a RepMgr-specific user and database.

sudo -u postgres createuser -s repmgr -p <CUSTOM_PORT>
sudo -u postgres createdb repmgr -O repmgr -p <CUSTOM_PORT>

Step 9: Configure pg_hba.conf to allow RepMgr access on the standby servers

File location:

/etc/postgresql/<VERSION>/<CLUSTER_NAME>/pg_hba.conf

Add entries for primary and standby nodes:

host repmgr <REPLICATION_USER> <PRIMARY_NODE_IP>/32  md5
host repmgr <REPLICATION_USER> <STANDBY_NODE_IP>/32 md5

Step 10: Create RepMgr Configuration Files

Prepare a RepMgr configuration file on each node.

Master Node Configuration

File: /etc/postgresql/repmgr/<CLUSTER_NAME>.conf

node_id=1
node_name='master-node'
conninfo='host=<MASTER_IP> port=<CUSTOM_PORT> user=repmgr dbname=repmgr connect_timeout=2'
data_directory='<DATA_PATH>'
pg_bindir='/usr/lib/postgresql/<VERSION>/bin/'
log_file='/var/log/repmgr/<CLUSTER_NAME>_repmgr.log'

Standby Node Configuration

File: /etc/postgresql/repmgr/<CLUSTER_NAME>.conf

node_id=2
node_name='standby-node'
conninfo='host=<STANDBY_IP> port=<CUSTOM_PORT> user=repmgr dbname=repmgr connect_timeout=10'
data_directory='<DATA_PATH>'
pg_bindir='/usr/lib/postgresql/<VERSION>/bin/'
log_file='/var/log/repmgr/<CLUSTER_NAME>_repmgr.log'
use_replication_slots=true
tablespace_mapping='/source_path=/target_path'

Step 11: Register the Primary Node in RepMgr

sudo -u postgres /usr/lib/postgresql/<VERSION>/bin/repmgr -f /etc/postgresql/repmgr/<CLUSTER_NAME>.conf cluster show

Step 12: Verify Cluster Status with RepMgr

sudo -u postgres /usr/lib/postgresql/<VERSION>/bin/repmgr -f /etc/postgresql/repmgr/<CLUSTER_NAME>.conf cluster show

Step 13: Perform a Dry Run for the Standby Clone

sudo -u postgres /usr/lib/postgresql/<VERSION>/bin/repmgr -h <MASTER_IP> -p <CUSTOM_PORT> -U repmgr -d repmgr -f /etc/postgresql/repmgr/<CLUSTER_NAME>.conf standby clone --dry-run

Step 14: Clone the Primary Node on the Standby Node

sudo -u postgres /usr/lib/postgresql/<VERSION>/bin/repmgr -h <MASTER_IP> -p <CUSTOM_PORT> -U repmgr -d repmgr -f /etc/postgresql/repmgr/<CLUSTER_NAME>.conf standby clone

Step 15: Register the Standby Node with RepMgr

sudo -u postgres /usr/lib/postgresql/<VERSION>/bin/repmgr -f /etc/postgresql/repmgr/<CLUSTER_NAME>.conf standby register

Step 16: Final Verification

sudo -u postgres /usr/lib/postgresql/<VERSION>/bin/repmgr -f /etc/postgresql/repmgr/<CLUSTER_NAME>.conf cluster show

Conclusion

This setup enables high availability and seamless failover for PostgreSQL streaming replication. Adapt these steps based on your PostgreSQL version and system requirements.

Also, feel free to share this guide with your network on LinkedIn or reach out for further discussion on streaming replication best practices.