postgres
The view that tells you whether your background writer is keeping up, whether your checkpoints are spread out properly, and whether your backends are doing write work they should not be doing.
postgres
The view that tells you how far behind each replica is, in three different units, with the asymmetric reading discipline that makes the difference between catching problems early and finding out at failover.
postgres
The tool that takes a binary copy of a running cluster while writes continue. The right primitive for point-in-time recovery and standby provisioning, but the documentation underweights the operational details that matter.
postgres
The file that decides who can connect, from where, and how they prove who they are. Most production incidents involving Postgres auth come from misreading the order of evaluation.
postgres
One per-database row, two dozen counters, and most of the operational signals you actually need. pg_stat_database is the highest-leverage observability view Postgres ships.
postgres
Bad query plans almost always trace back to stale or insufficient statistics. ANALYZE is one of the simplest Postgres operations and one of the most overlooked.
postgres
pg_dump's default settings are designed for small databases and reliability. On large databases they produce backups that take longer than they need to and restores that take orders of magnitude longer than they could.
postgres
VACUUM and CREATE INDEX and COPY and ANALYZE and CLUSTER and base backup can all run for hours. Postgres has dedicated progress views that tell you exactly where each one is in real time. Most teams discover them during the second incident.
engineering
Most production Postgres operational signals live in pg_stat_user_tables. The view is a per-table snapshot of insert/update/delete activity, sequential vs index scan counts, dead-tuple accumulation, and autovacuum history. Learning to read it is the difference between operating Postgres and react...
engineering
JSONB columns let you store flexible data, but the moment you need to search inside them, the default B-tree index is useless. GIN indexes solve the search problem at a cost most teams underestimate. The trade-offs deserve the same kind of attention you would give a schema migration.
engineering
The FILTER clause turns multi-pass conditional-aggregate queries into single-pass scans. Most teams writing 2026 Postgres code still reach for CASE WHEN inside SUM. There is a cleaner way that has been standard SQL since 2003.
engineering
Postgres has had COMMENT ON since version 7 in 2002. Almost no team uses it. The cost is one extra line per object and the value is that documentation lives where the schema lives and survives every refactor.