Welcome to pg_auto_failover’s documentation!

Contents:

• Introduction to pg_auto_failover
  • Single Standby Architecture
  • Multiple Standby Architecture
  • Multiple Standbys Architecture with 3 standby nodes, one async
• Main pg_autoctl commands
  • Command line environment, configuration files, etc
  • Creating Postgres Nodes
  • Inspecting nodes
  • Inspecting and Editing Replication Settings
  • Implementing Maintenance Operations
  • Manual failover, switchover, and promotions
  • What’s next?
• pg_auto_failover Tutorial
  • Create virtual network
  • Install the “pg_autoctl” executable
  • Run a monitor
  • Bring up the nodes
  • Node communication
  • Watch the replication
  • Cause a failover
  • Cause a node failure
  • Resurrect node B
• Architecture Basics
  • The pg_auto_failover Monitor
  • pg_auto_failover Glossary
    • Monitor
    • Formation
    • Group
    • Keeper
    • Node
    • State
  • Client-side HA
  • Monitoring protocol
  • Synchronous vs. asynchronous replication
  • Node recovery
• Multi-node Architectures
  • Architectures with two standby nodes
  • Replication Settings and Postgres Architectures
    • Number Sync Standbys
    • Replication Quorum
    • Candidate Priority
    • Auditing replication settings
  • Sample architectures with three standby nodes
  • Sample architecture with three standby nodes, one async
• Failover State Machine
  • Introduction
  • Example of state transitions in a new cluster
  • State reference
    • Init
    • Single
    • Wait_primary
    • Join_primary
    • Primary
    • Wait_standby
    • Catchingup
    • Secondary
    • Maintenance
    • Prepare_maintenance
    • Wait_maintenance
    • Draining
    • Demoted
    • Demote_timeout
    • Stop_replication
    • Prepare_promotion
    • Report_LSN
    • Fast_forward
  • Failover logic
  • pg_auto_failover keeper’s State Machine
• Failover and Fault Tolerance
  • Unhealthy Nodes
  • Network Partitions
  • Failure handling and network partition detection
• Installing pg_auto_failover
  • Ubuntu or Debian
    • Quick install
    • Manual Installation
  • Fedora, CentOS, or Red Hat
    • Quick install
    • Manual installation
  • Installing a pgautofailover Systemd unit
• Security settings for pg_auto_failover
  • Postgres HBA rules
  • The trust security model
  • Authentication with passwords
  • Encryption of network communications
  • Using your own SSL certificates
  • SSL Certificates Authentication
  • Postgres HBA provisioning
  • Enable SSL connections on an existing setup
• pg_autoctl commands reference
  • pg_autoctl command list
  • pg_auto_failover Monitor
    • pg_autoctl create monitor
    • pg_autoctl run
    • pg_autoctl create formation
    • pg_autoctl drop formation
  • pg_autoctl show command
    • pg_autoctl show uri
    • pg_autoctl show events
    • pg_autoctl show state
    • pg_autoctl show file
    • pg_autoctl show systemd
  • pg_auto_failover Postgres Node Initialization
    • pg_autoctl run
  • Replication Settings
    • pg_autoctl get formation settings
    • pg_autoctl get formation number-sync-standbys
    • pg_autoctl set formation number-sync-standbys
    • pg_autoctl get node replication-quorum
    • pg_autoctl set node replication-quorum
    • pg_autoctl get node candidate-priority
    • pg_autoctl set node candidate-priority
  • Configuration and State Files
    • Sample configuration file
    • Editing pg_autoctl configuration
    • Sample state file
    • Init State File
    • Sample PID file
  • Removing a node from the monitor
  • PG_AUTOCTL_DEBUG commands
• Configuring pg_auto_failover
  • pg_auto_failover Monitor
  • pg_auto_failover Keeper Service
• Operating pg_auto_failover
  • Deployment
  • Provisioning
  • Upgrading pg_auto_failover, from versions 1.4 onwward
  • Upgrading from previous pg_auto_failover versions
  • Cluster Management and Operations
    • Maintenance of a secondary node
    • Maintenance of a primary node
    • Triggering a failover
    • Implementing a controlled switchover
  • Current state, last events
  • Monitoring pg_auto_failover in Production
  • Trouble-Shooting Guide
• Frequently Asked Questions
  • The secondary is blocked in the CATCHING_UP state, what should I do?
  • Should I read the logs? Where are the logs?
  • The state of the system is blocked, what should I do?
  • The monitor is a SPOF in pg_auto_failover design, how should we handle that?
    • Consequences of the monitor being unavailable
    • The monitor node can be brought up again without data having been lost
    • The monitor node can only be built from scratch again