Neues vom PostgreSQL Planet

You have probably read a dozen tutorials on setting up PostgreSQL High Availability (HA). On paper, it looks simple: spin up a primary instance, spin up a standby, and let them replicate.

From October 20-23, the global Postgres community will descend upon the beautiful Palacio de Congresos in Valencia, Spain, for PGConf EU. I compiled an overview of the sessions and community events EDB colleagues are running.

Partitioning gets sold as a performance switch. You flip it on a big table, and the slow queries get fast. Most of the time, that is exactly what happens. But there is a version of this where you do all the work, split a huge table into clean partitions, and the slow queries stay exactly as slow as they were. The table is partitioned. Nothing got better.

Every PostgreSQL connection since 8.0 has heard the server announce `integer_datetimes`—a parameter that's had only one possible value since PostgreSQL 10.

The PostGIS Team is pleased to release PostGIS 3.7.0beta2! Best Served with PostgreSQL 19 Beta2 and GEOS 3.15.0beta2.

This version requires PostgreSQL 14 - 19beta2, GEOS 3.10 or higher, and Proj 6.1+. To take advantage of all features, GEOS 3.15+ is needed. To take advantage of all SFCGAL features SFCGAL 2.3.0+ is needed.

Enterprise AI doesn’t fail because strategy was wrong. Or because the platform wasn’t ready. Or because the execution layer broke.

It fails because organizations treat those as three different conversations — run by three different teams, on three different timelines — instead of what they actually are: three altitudes of the same question.

After becoming a PostgreSQL ACE, Zhang Chen explains why he uses the title “PostgreSQL Database Recovery Expert” and how PDU evolved from offline extraction and WAL mining to fragment scanning and recovery without a data dictionary.

Recover from catalog index corruption by sequentially scanning system tables instead of using indexes.

If you maintain packages for any length of time, you know the real risk isn't building them — it's missing an upstream release. A CVE fix ships quietly on a Tuesday (or Friday!), a minor version bumps without fanfare, and if nobody's watching, it sits unnoticed until a user files a bug report asking why the repo is stale.

Release Radar, a self-hosted project written by Claude and me, is built to close exactly that gap.

** What it does

Before diving into this final post covering new Postgres 19 features, I just want to say it's been a wild ride. Postgres 19 has been a veritable treasure trove of enhancements, perhaps more than any previous release; or maybe that's just my perception. Usually I just skim through the release notes and nod along, sometimes jotting down things that look interesting for later study. Maybe calling out each element that caught my eye was the right thing to do, to really show how far Postgres has come since the last release, rather than simply accepting the status quo.

The Talk Selection Committee of PGConf.EU 2026 met to finalize the conference schedule:

When a standby refuses to replay WAL that references pages that don't exist, `ignore_invalid_pages` converts the PANIC into a WARNING—but only as a last resort…

Build and verify PostgreSQL row-level security for AI agents with non-owner roles, FORCE RLS, explicit policies, controlled writes, and denial tests.

It can looks like (2026-08-06 17:44:32) postgres=# \luacode Enter code to be copied followed by a newline. End with a backslash and a period on a line by itself, or an EOF signal. >> function x(n) >> return n + 10 >> end >> \. (2026-08-06 17:45:32) postgres=# \luacode Enter code to be copied followed by a newline. End with a backslash and a period on a line by itself, or an EOF signal. >> print (x(10)) >> \. 20

Why AI Reliability Is an Execution Problem, Not a Model Problem

The most dangerous AI decision isn’t the wrong one. It’s the one nobody can explain afterward.

PostgreSQL is already the default database for agentic AI. That question is settled. But the more agentic your workloads get, the more your database needs to do. Models and workflows flood it with signals, state, memory, and checkpoints, and most teams just absorb the flood, treating PostgreSQL like a parking lot rather than a compute layer. The people building these systems are AI engineers, not database people.

CloudNativePG lets the ClusterImageCatalog carry extension images alongside the operand, a capability every currently supported release already has, so a Cluster manifest only needs to name an extension and nothing else. This recipe deploys the community’s extension catalog and shows the operator resolving pgvector’s image, paths and dependencies from a single, versioned source of truth per PostgreSQL major version.