PostgreSQL error: database does not exist

SQLSTATE 3D000 (invalid_catalog_name) · Updated July 9, 2026

FATAL:  database "gianluca" does not exist

You connected to the cluster successfully, but asked for a database that isn't there. When the "missing database" in the message is your own username, nothing is broken: psql simply defaults the database name to the user name when you don't specify one.

What this error means

A PostgreSQL server (cluster) hosts multiple databases; every connection targets exactly one. If the requested name doesn't exist on the cluster you actually reached, you get this FATAL. Both halves of that sentence cause confusion:

• The requested name may not be what you think — psql and many tools default it (to the user name), and environment variables like PGDATABASE override quietly.
• The cluster you actually reached may not be the one with your data: port 5432 vs 5433, host machine vs Docker container, local vs remote.

Common causes

• psql without -d: database name defaults to the user name — the classic database "gianluca" does not exist.
• The database hasn't been created yet in this environment (fresh setup, new CI job, recreated container without its init scripts).
• Wrong cluster: connection string points at another port/host where that database genuinely doesn't exist.
• Case or typo: database names are identifiers — created quoted as "AppDb", it must be requested with that exact case.

How to diagnose it

# Cosa c'è davvero su questo cluster:
psql -h HOST -p 5432 -U USER -d postgres -c '\l'

Connecting to the always-present postgres maintenance database and listing what exists answers both questions at once: whether you're on the right cluster, and what the database is actually called.

How to fix it

• Name the database explicitly: psql -d appdb, or fix the connection string / PGDATABASE.
• Create it if it's genuinely missing:

  CREATE DATABASE appdb OWNER app_user;

• Wrong cluster → fix host/port; on Debian-family systems pg_lsclusters lists local clusters and their ports.