RDS PostgreSQL

Based on the RDS PostgreSQL docker image this Terraform module provides a ready to use PostgreSQL server that is backed up to a S3 compatible object store.

The module supports two backup methods either local storage (Hetzner Cloud Volume) or a S3 compatible object store. To avoid data loss it is not possible to run the database without one of the backup methods enabled.

Design Goals

Acknowledging that there are a lot of available solutions to run PostgreSQL in the cloud it is important to emphasize the design goals of this solution to be able to make an informed decision on if and how to use it

• First and foremost goal is to preserve the data stored in the PostgreSQL instance. This is ensured by an extensive set of full integration tests replicating the intended usage scenarios
• Reduced complexity through single node operations. Running a PostgreSQL server with replication or in multi master comes with its own set of diverse failure scenarios and edge-cases. This solution aims at workloads that can easily handled by a single node and if in doubt favours MTTR over MTBF
• Reduced complexity through less moving parts. Features like autoscaling, automatic recovery after zone outages, user interfaces, etc. like they are provided by the Zalando PostgreSQL Operator provide a lot of value but also add multiple layers between the VM and the actual PostgreSQL instance. This solutions explicitly tries to avoid these features trying to keep the stack as simple as running a docker container on a VM.

Usage

data "aws_s3_bucket" "backup" {
  bucket = "test-rds-postgresql-backup"
}

data "hcloud_volume" "data" {
  name = "rds-postgresql-data"
}

resource "tls_private_key" "ssh_key" {
  algorithm = "RSA"
  rsa_bits  = 4096
}

resource "hcloud_ssh_key" "ssh_key" {
  name       = "rds-postgresql"
  public_key = tls_private_key.ssh_key.public_key_openssh
}

module "rds-postgresql" {
  source = "https://github.com/pellepelster/solidblocks/releases/download/v0.4.12/blcks-terraform-rds-postgresql-hetzner-v0.4.12.zip"

  name     = "rds-postgresql"
  location = var.hetzner_location

  ssh_keys = [hcloud_ssh_key.ssh_key.id]

  data_volume = data.hcloud_volume.data.id

  backup_s3_bucket     = data.aws_s3_bucket.backup.id
  backup_s3_access_key = var.backup_s3_access_key
  backup_s3_secret_key = var.backup_s3_secret_key

  databases = [
    { id : "database1", user : "user1", password : "password1" }
  ]
}

For more usage examples see Extended Usage

Operations

To access the created instance use one of SSH keys that was provided via ssh_keys. All linux system resources include the provided name to make them distinguishable.

Show Logs

journalctl -u rds@rds-postgresql-${name}.service

Start/Stop Database

systemctl [start|stop] rds@rds-postgresql-${name}.service 

Maintenance Mode

In case of errors or the need for manual intervention a maintenance mode is available. Triggering the maintenance mode will set up the VM like it would be set up for database startup, but without actually starting the database. This allows to exec into the container to debug issues.

main.tf

module "rds-postgresql" {
  source  = "pellepelster/solidblocks-rds-postgresql/hcloud"
  # [...]
  mode = "maintenance"
  # [...]
}

docker exec -ti rds-postgresql-${name}_postgresql /bin/bash

Terraform Module

Requirements

Providers

Resources

Inputs

Outputs