Neues vom PostgreSQL Planet

Google Summer of Code is back for 2026! We’re celebrating the 22nd year of this incredible program that has brought countless talented developers into the open-source world. Please take a moment to review Google’s announcement and familiarize yourself with what makes this year special.

PostgreSQL 19 changes this week logical replication logical decoding enabled dynamically based on logical slot presence and does not require wal_level to be set to logical PostgreSQL 18 articles Instant database clones with PostgreSQL 18 (2025-12-22) - Radim Marek

more...

© Laurenz Albe 2025

Starting with arrays in PostgreSQL is as simple as declaring a column as integer[], inserting some values, and you are done.

Or building the array on the fly.

Happy New Year! I'm excited to announce release 0.9 of pg_statviz, the minimalist extension and utility pair for time series analysis and visualization of PostgreSQL internal statistics.

This is a significant feature release that expands the scope of analysis to include several new modules and a visualization update:

Why static batch size configuration breaks down in real world networks and how automatic batch size tuning improves snapshot throughput.

Meet the mentee: Anushka Saxena worked with the project maintainers on transforming the documentation for CloudNativePG, as part of the LFX mentorship program.

At first glance, the idea that an idle session could lead to a transaction wraparound might seem counterintuitive. Most PostgreSQL users are familiar with how long-running queries or open transactions can cause table bloat and wraparound risks by pinning the xmin horizon, which prevents autovacuum from reclaiming dead tuples and Transaction IDs.

PostgreSQL 19 changes this week cumulative statistics support for reading/writing auxiliary data from/to disk added pg_createsubscriber support for handling existing publications added pg_sync_replication_slots()  retry logic added PostgreSQL 18 articles PostgreSQL 18 Asynchronous Disk I/O - Deep Dive Into Implementation (2025-12-15) - Josef Machytka / Credativ

more...

When the database won't start and there's no backup available, most teams assume the game is over. But that's not true—the data is still on disk, it just lost the ability to speak. PostgreSQL data files remain transparent and readable, but here's the problem - without the data dictionary, you cannot understand what those bytes actually mean.

When people first hear "instant recovery," they often assume it is a replacement for backup—or worse, a risky shortcut only experts should attempt. But the truth is exactly the opposite. Instant recovery does not challenge any design boundary of PostgreSQL.

PostgreSQL 19 changes this week ALTER TABLE ALTER TABLE ... SPLIT PARTITION ... syntax added ALTER TABLE ... MERGE PARTITIONS ... syntax added pg_stat_progress_analyze column started_by added pg_stat_progress_vacuum columns mode and started_by added vacuumdb option --dry-run added PostgreSQL 18 articles Postgres 18 New Default for Data Checksums and How to Deal with Upgrades (2025-12-11) - Greg Sabino Mullane / Crunchy Data

more...

As we wrap up 2025 and prepare for another exciting year of 2026, HexaCluster is back again with our annual Summary of PostgreSQL in 2025, continuing the tradition we set in 2024.

Working in development, there is a common challenge: how to attach custom metadata to database objects without modifying Postgres's core code. In this article, I briefly demonstrate a practical solution using Postgres's SECURITY LABELS mechanism to implement custom properties that are transactional, properly linked to database objects, and work with standard Postgres operations.

Back in September 2025, David Fetter asked on IRC, about a tool to quickly load CSV to database. One that would require minimal configuration, will try to figure out as much as possible on its own. I thought that it would be great idea. Plus, I'm trying to learn more JavaScript / Node, so figured … Continue reading "Quick and dirty loading of CSV files"

On December 10th, 2025, the PostgreSQL December Meetup in Frankfurt (Main) took place. We had two speakers, and a nice dinner. This Meetup was organized around the IT-Tage, and the Meetup organizers had booked a reeting toom in the Scandic Frankfurt Museumsufer. About 15 minutes walking from the conference, and unfortunately it was raining that evening. Meeting room Dirk Aumueller: Running PostgreSQL with Podman, Quadlet & Systemd Dirk spoke about how to run PostgreSQL inside a Podman Quadlet which is managed by systemd.

As you maybe know, some time ago I made paste service, mostly to use for queries, or related text to share on IRC. One part of it is that it also has pretty printer of provided queries. Recently I realized that in case of complex join conditions, the output is, well, sub-optimal. For example: SELECT … Continue reading "Small improvement for pretty-printing in paste.depesz.com"

One of things people learn is that adding indexes isn't free. All write operations (insert, update, delete) will be slower – well, they have to update index. But realistically – how much slower? Full tests should involve lots of operations, on realistic data, but I just wanted to see some basic info. So I figured … Continue reading "What is index overhead on writes?"

I am writing this post over the weekend but scheduling it to be published on Tuesday, after the PG DATA CfP closes, because I do not want to distract anyone, including myself, from the submission process.

A couple of months ago, I created a placeholder in my GitHub, promising to publish pg_acm before the end of the year. The actual day I pushed the initial commit was January 3, but it still counts, right? At least, it happened before the first Monday of 2026!

I do a fair amount of benchmarks as part of development, both on my own patches and while reviewing patches by others. That often requires dealing with noise, particularly for small optimizations. Here’s an overview of ways I use to filter out random variations / noise.

Most of the time it’s easy - the benefits are large and obvious. Great! But sometimes we need to care about cases when the changes are small (think less than 5%).