Mountpoints and the Single PostgreSQL Server
PostgreSQL databases live in directories, and naming them wisely saves you headaches during major version upgrades.
Why PostgreSQL is Terrible? at Nordic PGDay
Slides from "Why PostgreSQL is Terrible" talk at Nordic PGDay are online. Discover what makes PostgreSQL unexpectedly great through its apparent flaws.
A Replication Cheat-Sheet
PostgreSQL replication overwhelms with options?
Commitment Issues
When a network disconnect interrupts your COMMIT, did the transaction really go through?
Streaming replication stopped? One more thing to check.
Streaming replication froze with thousands of locks on non-existent tables.
“Humans do not have a primary key” at PGConf.EU
Database designers often assume every entity needs a primary key. Humans are messier than that.
“Humans do not have a primary key” at PGOpen SV
Humans don't have primary keys. Learn why treating people as database records breaks down when you account for names, identity, and real-world complexity.
A brief reminder about casting TIMESTAMPTZ
Casting TIMESTAMPTZ to DATE gives different results depending on your client time zone—a subtle bug that can make identical queries return different dates.
Blog Moved.
The blog has moved to a new server. Here's what changed and what stayed the same.
shared_buffers is not a sensitive setting
Most shared_buffers tuning advice is overthinking it. Set it to 25% of system memory (max 32GB) and move on—the difference between 24GB and 32GB won't matter.