Upgrading

While PGD and Postgres are closely related, they’re separate products with separate upgrade paths. This section covers how to upgrade both PGD and Postgres.

Upgrading PGD

EDB Postgres Distributed is a flexible platform. This means that your upgrade path depends largely on how you installed PGD.

• ** Upgrading PGD clusters manually ** — If you manually installed and configured your PGD cluster, you can move a cluster between versions, both minor and major.

• ** Supported PGD upgrade paths ** — Several supported upgrade paths are available.

• ** PGD compatibility ** — If you’re upgrading from PGD 3.x or 4.x to PGD 5.x or later, you need to understand the compatibility changes between versions.

• ** Upgrading application schemas ** — How to apply DDL changes across a live cluster one node at a time, using rolling application schema upgrades and conflict resolver configuration.

Upgrading Postgres or Postgres and PGD major versions

• **

In-place Postgres major version upgrades ** — How to use pgd node upgrade to manually

upgrade the Postgres version or Postgres and PGD major version on one or more nodes.

• ** Performing a Postgres major version rolling upgrade on a PGD cluster ** — How to perform a major version upgrade of Postgres on a cluster.

• Upgrading PGD clusters manually
  • Upgrade planning
    • Rolling upgrade considerations
    • Rolling server software upgrades
    • Rolling upgrade using node join
    • Upgrading a CAMO-enabled cluster
  • Upgrade preparation
  • Server software upgrade
    • BDR extension upgrade
    • Fence the node
    • Stop Postgres
    • Upgrade packages
    • Start Postgres
    • Unfence the node
    • Minor version Postgres upgrade
    • Major version Postgres upgrade
  • Upgrade check and validation
  • PGD 5 - Moving from PGD Proxy to Connection Manager
• Supported PGD upgrade paths
  • Upgrading to version Version 6.4
• Compatibility changes
  • Summary
  • Connection Manager
  • DDL Locking changes
  • Join/part behavior
    • Function and node state changes
    • Sequences conversion
    • Parting a node drops it
  • Administration function changes
  • Other
    • Replica identity
    • Run on (all) writes
    • General UI improvements
  • Changes to defaults
    • Global lock timeout
    • Auto sync
• Upgrading application schemas
  • Preparing for a rolling application schema upgrade
    • Adding tables
    • Adding columns
    • Removing columns
    • Changing column types
  • Performing a rolling application schema upgrade
  • Using transform triggers to bridge schema differences
• In-place Postgres or Postgres and PGD major version upgrades
  • Terminology
  • Precautions
  • Usage
    • pgd node upgrade command-line
    • Synopsis
    • Options
    • Environment variables
    • Example
    • Steps performed
    • PGD Postgres checks
    • pg_upgrade steps
    • PGD post-upgrade steps
• Performing a Postgres major version rolling upgrade on a PGD cluster
  • Upgrading Postgres major versions
    • Prepare the upgrade
    • Perform the upgrade on each node
  • Worked example: Upgrade PGD 4 to PGD 6
    • Overview
    • Confirm the harp-proxy leader
    • Fence the node
    • Stop the Postgres service
    • Stop HARP manager
    • Remove and install packages
    • Start the Postgres service
    • Start HARP manager
    • Unfence the node
    • Repeat steps for all nodes
    • Confirm cluster version
    • Confirm SCRAM hashes
    • Enable routing
    • Switch to Connection Manager
    • Stop harp manager and proxy services.
  • Worked example: Upgrade PGD 5 to PGD 6
  • Prerequisites
    • Install packages for the new server and PGD
  • Pre-upgrade steps
    • Version Check
    • Move to Connection Manager
    • Write leader node verification
    • Fence the node
    • Initialize the new Postgres instance
    • Migrate configuration to the new Postgres version
    • Execute the upgrade.
  • Post-upgrade steps
    • Update the Postgres service file
    • Start the postgres service
    • Verify the upgraded cluster versions
    • Unfence the node
    • Verify the Connection Manager is working
    • Clean up and vacuum analyze
    • Upgrade the remaining nodes
    • Verify the final state of the cluster
    • Verify the Connection Manager