6 min read
Deadlock Prevention in Postgres: Design It Out, Do Not Just Retry
Deadlocks are the symptom of a design that allows two transactions to hold each other's locks. The fix is rarely retry logic. Here is what to look for instead.
PostgreSQL Field Notes
Notes for the problems that show up after launch: bad plans, awkward migrations, index debt, vacuum pressure, replica lag, and the small decisions that make PostgreSQL easier to operate.
PostgreSQL Topic Hubs
PostgreSQL Tools
Tool Comparisons
5 min read
ALTER TABLE Locks in Postgres: The Reference I Wish I Had Five Years Ago
Some ALTER TABLE operations are instant. Others lock the table for the duration. The Postgres docs hide this in subsections. Here is the cheat sheet.
6 min read
VACUUM FREEZE in Postgres: The 4 a.m. Page You Can Prevent
Transaction ID wraparound is rare, dramatic, and entirely avoidable. Here is what FREEZE actually does, and the autovacuum settings that keep you out of single-user mode.
6 min read
Measuring Table Bloat in Postgres: The Question, Answered Honestly
The popular bloat estimation queries are off by 10-30% under typical workloads. Here is what pgstattuple actually measures, when the approximate variant is enough, and the cleanup decisions that follow.
12 min read
PostgreSQL Memory Tuning: shared_buffers, work_mem, and Cache
Memory tuning is mostly about budgets. shared_buffers, OS cache, per-query work_mem, maintenance work, and connection count all spend the same RAM.
7 min read
When the Postgres Disk Fills: A Playbook From the 2 a.m. End of It
Disk-full on a Postgres server is rarely just one thing. WAL, temp files, logs, and delayed cleanup arrive together. The recovery is mostly about what you prepared before.
6 min read
Postgres Connection Storms: When the Pool Becomes the Outage
Every connection storm I have seen looks like a database problem and is actually an application or pool problem. Here is how to tell the difference and stop blaming Postgres.
7 min read
Point-in-Time Recovery in Postgres: The Capability You Hope You Never Use
PITR lets you restore the database to any moment within your retention window. The feature is well-documented; the operational reality is messier.
6 min read
Postgres Restore Drills: The Backup You Have Not Tested Is Not a Backup
Most teams have backups. Most teams have never restored from one. The first time you test, you find out the backup was missing something.
6 min read
pg_dump vs Physical Backup: Picking the Right Postgres Backup Strategy
Logical and physical backups solve different problems. Most teams need both, but only have one. Here is the actual decision framework.
5 min read
WAL Archive Retention: How Long to Keep WAL and How to Clean Up Safely
WAL accumulates fast. The retention policy is a tradeoff between PITR window and storage cost. Here is how I think about it.
12 min read
PostgreSQL Connection Pooling: PgBouncer Without Surprises
Pooling is not just raising throughput. It is controlling backend count, transaction shape, prepared statements, and failure behavior under load.