Deploy PostgreSQL HA with Repmgr on Railway

Deploy a PostgreSQL HA Cluster with replication manager by EDB, for a more highly available and redundant data layer.

How it works

This template deploys 4 services -

(3) PostgreSQL nodes
(1) Pgpool-II service

The PostgreSQL nodes are automatically configured with repmgr and joined to a cluster. Since Postgres doesn't include a native failover, repmgr is included.

Pgpool-II is a PostgreSQL proxy which can also handle failovers when they occur in the cluster.

Inter-node communication

All communication between nodes is done over the private network.

Configuration Tuning

The PostgreSQL nodes and Pgpool-II are deployed from bitnami images. As such, you can leverage the customizable environment variables available in both -

Connecting to the Cluster

Within the service variables for the Pgpool service, you will find connection strings named DATABASE_URL and DATABASE_PUBLIC_URL. These can be used for connecting to the cluster via the private or public network.

We strongly encourage you to use the private connection string defined in DATABASE_URL to avoid egress fees.

Deployment Regions

By default, each of the 3 Postgres nodes are deployed into US-West. You can move them to your preferred region(s) once deployed into your project, from the service settings.

From a disaster preparedness standpoint, it is common practice to deploy the nodes into different regions.

Service Source

PostgreSQL nodes- As of now, the PostgreSQL nodes deploy from a forked repository in the railwayapp-templates Github while we await a PR review by bitnami. Once our PR has been merged and the image published via Bitnami, we will update the source

Pgpool-II - Pgpool deploys directly from the bitnami/pgpool image in Docker Hub.