README.md

patroni

Table of Contents

1. Description
2. Setup - The basics of getting started with patroni
   • What patroni affects
   • Setup requirements
   • Beginning with patroni
3. Usage - Configuration options and additional functionality
4. Reference - An under-the-hood peek at what the module is doing and how
5. Limitations - OS compatibility, etc.
6. Development - Guide for contributing to the module
7. License

Description

This module sets up a Patroni instance, which provides seemless replication for PostgreSQL, allowing you to run a load balanced and highly available PostgreSQL service. It is one of many options for HA with PostgreSQL, so please take a look at the myriad of other options to make sure you pick the one that is right for your environment.

This module alone is not enough to run a fully HA and replicated service. Please read up on your options at Patroni's GitHub Project. In our case, we use haproxy, using puppetlabs's haproxxy module, and etcd, using Tailored Automation's etcd module.

This module was originally written by Jadestorm. Thank you!!!

Setup

What patroni affects

The patroni module sets up the following:

• Install necessary PostgreSQL packages
• Installs Patroni via Pip or package
• Sets up a systemd based service for Patroni
• Manages Patroni's configuration

Setup Requirements

It is very important that you read up on how Patroni works, as you will also need a variety of other components to accomplish anything useful with Patroni.

If not installing via Pip and setting $install_method => 'package', you also need to make sure the patroni package is available somewhere. For RPM based systems, you can get the package from here.

Beginning with patroni

A bare minimum configuration might be:

class { 'patroni':
  scope => 'mycluster',
}

If you specify patroni::pgsql_parameters at multiple levels in Hiera and would like to merge them, add the following to your common.yaml (or lowest priority Hiera level).

lookup_options:
  patroni::pgsql_parameters:
    merge: deep

This assumes you have taken care of all of the rest of the components needed for Patroni.

Usage

Below is a full example:

# First PostgreSQL server
node pg1 {
  class { 'etcd':
    etcd_name                   => ${::hostname},
    listen_client_urls          => 'http://0.0.0.0:2379',
    advertise_client_urls       => "http://${::fqdn}:2379",
    listen_peer_urls            => 'http://0.0.0.0:2380',
    initial_advertise_peer_urls => "http://${::fqdn}:2380",
    initial_cluster             => [
      'pgarb=http://pgarb.example.org:2380',
      'pg1=http://pg1.example.org:2380',
      'pg2=http://pg2.example.org:2380',
    ],
    initial_cluster_state       => 'existing',
  }

  class { 'patroni':
    scope                   => 'mycluster',
    use_etcd                => true,
    pgsql_connect_address   => "${::fqdn}:5432",
    restapi_connect_address => "${::fqdn}:8008",
    pgsql_bin_dir           => '/usr/pgsql-9.6/bin',
    pgsql_data_dir          => '/var/lib/pgsql/9.6/data',
    pgsql_pgpass_path       => '/var/lib/pgsql/pgpass',
    pgsql_parameters        => {
      'max_connections' => 5000,
    },
    pgsql_pg_hba            => [
      'host all all 0.0.0.0/0 md5',
      'host replication rep_user 0.0.0.0/0 md5',
    ],
    superuser_username      => 'postgres',
    superuser_password      => 'somepassword',
    replication_username    => 'rep_user',
    replication_password    => 'someotherpassword',
  }
}
# Second PostgreSQL server
node pg2 {
  class { 'etcd':
    etcd_name                   => ${::hostname},
    listen_client_urls          => 'http://0.0.0.0:2379',
    advertise_client_urls       => "http://${::fqdn}:2379",
    listen_peer_urls            => 'http://0.0.0.0:2380',
    initial_advertise_peer_urls => "http://${::fqdn}:2380",
    initial_cluster             => [
      'pgarb=http://pgarb.example.org:2380',
      'pg1=http://pg1.example.org:2380',
      'pg2=http://pg2.example.org:2380',
    ],
    initial_cluster_state       => 'existing',
  }

  class { 'patroni':
    scope                   => 'mycluster',
    use_etcd                => true,
    pgsql_connect_address   => "${::fqdn}:5432",
    restapi_connect_address => "${::fqdn}:8008",
    pgsql_bin_dir           => '/usr/pgsql-9.6/bin',
    pgsql_data_dir          => '/var/lib/pgsql/9.6/data',
    pgsql_pgpass_path       => '/var/lib/pgsql/pgpass',
    pgsql_parameters        => {
      'max_connections' => 5000,
    },
    pgsql_pg_hba            => [
      'host all all 0.0.0.0/0 md5',
      'host replication rep_user 0.0.0.0/0 md5',
    ],
    superuser_username      => 'postgres',
    superuser_password      => 'somepassword',
    replication_username    => 'rep_user',
    replication_password    => 'someotherpassword',
  }
}
# Simple etcd arbitrator node, meaning it serves no content of it's own, just helps keep quorum
node pgarb {
  class { 'etcd':
    etcd_name                   => ${::hostname},
    listen_client_urls          => 'http://0.0.0.0:2379',
    advertise_client_urls       => "http://${::fqdn}:2379",
    listen_peer_urls            => 'http://0.0.0.0:2380',
    initial_advertise_peer_urls => "http://${::fqdn}:2380",
    initial_cluster             => [
      'pgarb=http://pgarb.example.org:2380',
      'pg1=http://pg1.example.org:2380',
      'pg2=http://pg2.example.org:2380',
    ],
    initial_cluster_state       => 'existing',
  }
}

Reference

All of the Patroni settings I could find in the Patroni Settings Documentation are mapped to this module. However, I do not have experience with the bulk of those settings, so implementing them here was done as a best guess.

I also highly recommend checking out PostgreSQL High Availability Cookbook as it is a fantastic resource for wrapping your head around all of the options and has a great walkthrough for setting up Patroni.

Limitations

This module is currently only supported on RHEL and Debian based operating systems that support Systemd.

Development

See CONTRIBUTING.md

License

See LICENSE file.