Minimal Install

How to perform minimal install with HA PostgreSQL related components only

Pigsty has an entire infrastructure stack as an enclosure of HA PostgreSQL clusters, BUT it is viable to install only the PostgreSQL components without the rest of the stack. This is called a minimal installation.


Overview

The minimal installation focus on Pure HA-PostgreSQL Cluster, and it only installs essential components for this purpose.

There’s NO Infra modules, No monitoring, No local repo Just partial of NODE module, along with ETCD & PGSQL modules

Systemd Service Installed in this mode:

  • patroni: REQUIRED, bootstrap HA PostgreSQL cluster
  • etcd: REQUIRED, DCS for patroni
  • pgbouncer: OPTIONAL, connection pooler for postgres
  • vip-manager: OPTIONAL, if you want to use a L2 VIP bind to primary
  • haproxy: OPTIONAL, if you wish to auto-routing service
  • chronyd: OPTIONAL, if you wish to sync time with NTP server
  • tuned: OPTIONAL, manage node template and kernel parameters

You can turn-off the optional components, the only two essential components are patroni and etcd.


Configure

To perform a minimal installation, you need to disable some switches in the pigsty.yml config file:

all:

  children:
    infra: { hosts: { 10.10.10.10: { infra_seq: 1 } }, vars: { docker_enabled: true } }

    etcd:
      hosts:
        10.10.10.10: { etcd_seq:  1 }
        #10.10.10.11: { etcd_seq:  2 } # optional
        #10.10.10.12: { etcd_seq:  3 } # optional
      vars: { etcd_cluster: etcd }

    pg-meta:
      hosts:
        10.10.10.10: { pg_seq: 1, pg_role: primary }  # init one single-node pgsql cluster by default, with:
        #10.10.10.11: { pg_seq: 2, pg_role: replica } # optional replica : bin/pgsql-add pg-meta 10.10.10.11
        #10.10.10.12: { pg_seq: 3, pg_role: replica } # optional replica : bin/pgsql-add pg-meta 10.10.10.12
      vars:
        pg_cluster: pg-meta
        pg_users:         # define business users here: https://pigsty.io/docs/pgsql/user/
          - { name: dbuser_meta ,password: DBUser.Meta ,pgbouncer: true ,roles: [ dbrole_admin ] ,comment: pigsty default user }
        pg_databases:     # define business databases here: https://pigsty.io/docs/pgsql/db/
          - { name: meta ,comment: pigsty default database  }
        pg_hba_rules:     # define HBA rules here: https://pigsty.io/docs/pgsql/hba/#define-hba
          - { user: dbuser_meta , db: all ,addr: world ,auth: pwd ,title: 'allow default user world access with password (not a good idea!)' }
        node_crontab:     # define backup policy with crontab (full|diff|incr)
          - '00 01 * * * postgres /pg/bin/pg-backup full'
        #pg_vip_address: 10.10.10.2/24  # optional l2 vip address and netmask
        pg_extensions:    # define pg extensions (336 available): https://pigsty.io/docs/pgext/
          - postgis timescaledb pgvector

  vars:
    version: v3.0.2                   # pigsty version string
    admin_ip: 10.10.10.10             # admin node ip address
    region: default                   # upstream mirror region: default|china|europe
    node_tune: tiny                   # use tiny template for NODE  in demo environment
    pg_conf: tiny.yml                 # use tiny template for PGSQL in demo environment
    # minimal installation setup
    node_repo_modules: node,infra,pgsql
    nginx_enabled: false
    dns_enabled: false
    prometheus_enabled: false
    grafana_enabled: false
    pg_exporter_enabled: false
    pgbouncer_exporter_enabled: false
    pg_vip_enabled: false

And instead of using the install.yml playbook, use the pgsql-min.yml instead:

./pgsql-min.yml

Last modified 2024-09-07: bump version to v3.0.2 (48a585d)