All articles

Databases · · 7 min read

Upgrading Amazon RDS PostgreSQL 14 to 17: a practical runbook

By Rushit Chandegara, Founder & CTO, InfraCloudEdge

PostgreSQL 14 is approaching the end of standard support on Amazon RDS, after which AWS charges for Extended Support. Moving straight to PostgreSQL 17 skips two upgrade cycles, but a major version jump needs a plan. This is the runbook we use for production fleets.

1. Rehearse on a snapshot first

Restore a snapshot of production into a new instance and upgrade that copy. It tells you how long the upgrade takes, whether any extension blocks it, and gives your developers a PostgreSQL 17 database to test against without touching production.

2. Watch for the SSL default

This is the change that catches most teams. In the default parameter groups for RDS PostgreSQL 15 and later, rds.force_ssl is on, so the database refuses unencrypted connections. An application that connects with sslmode=disable, or an ORM configured the same way, will fail straight after the upgrade with an authentication or "no encryption" error.

  • Search every connection string and environment variable for sslmode before the upgrade.
  • The right fix is to connect with SSL (sslmode=require, or verify-full with the RDS CA bundle).
  • Only as a temporary measure, a custom parameter group with rds.force_ssl set to 0 restores the old behaviour.

3. Choose the upgrade method

An in-place major upgrade is simple but takes the database offline for the whole upgrade. RDS Blue/Green Deployments create an upgraded copy kept in sync through logical replication, and the switchover usually takes about a minute.

Blue/Green has rules of its own. Schema changes (DDL) are not replicated while the deployment is running, so freeze migrations for the window. Check the AWS documentation for the current list of limitations for your engine version before you rely on it.

4. Extensions and parameter groups

  • List installed extensions (SELECT * FROM pg_extension) and check each one is supported on PostgreSQL 17.
  • Run ALTER EXTENSION ... UPDATE after the upgrade where a newer version is available.
  • Create a PostgreSQL 17 parameter group that carries over your custom settings; the 14 group cannot be reused.

5. After the upgrade

  • Run ANALYZE on every database. Before PostgreSQL 18, optimizer statistics are not carried across a major upgrade, and queries can be slow until they are rebuilt.
  • Compare slow-query logs and Performance Insights with the week before.
  • Keep the old instance or a final snapshot until the application has run cleanly for a few days.

Order matters across a fleet

With several databases, upgrade the least critical first: development, then QA, then production, one at a time. Each step turns surprises into runbook lines before they reach production.

More articles