| From: | PostgresAI via PostgreSQL Announce <announce-noreply(at)postgresql(dot)org> |
|---|---|
| To: | PostgreSQL Announce <pgsql-announce(at)lists(dot)postgresql(dot)org> |
| Subject: | pg_ash v1 - Active Session History for PostgreSQL |
| Date: | 2026-04-03 09:06:09 |
| Message-ID: | 177520716916.3221089.1804726113216120208@wrigleys.postgresql.org |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-announce |
pg_ash samples `pg_stat_activity` once per second via `pg_cron`, stores encoded snapshots in partitioned tables, and provides 32 SQL functions for wait event analysis.
Pure SQL + PL/pgSQL. No C extension, no `shared_preload_libraries`, no restart required. Works anywhere `pg_cron` runs — managed providers (RDS, Cloud SQL, AlloyDB, Azure Database for PostgreSQL, Supabase, etc.), self-managed, Kubernetes, or any other deployment.
Storage: ~100-200 bytes per sample, ~30 MiB/day with `TRUNCATE`-based partition rotation (zero bloat).
Analysis functions include top waits, per-query wait profiles, wait event drill-down, time-range investigation, and timeline visualization. Functions are designed to chain naturally for iterative root cause analysis — by a human in psql or by AI tools such as Claude Code, OpenCode, or Codex.
PostgreSQL 14-18. Apache 2.0. CI tested.
[https://github.com/NikolayS/pg_ash](https://github.com/NikolayS/pg_ash)
| From | Date | Subject | |
|---|---|---|---|
| Previous Message | Gavin M. Roy via PostgreSQL Announce | 2026-04-03 09:03:54 | tree-sitter-postgres, libpgfmt, pgfmt, and libpgdump |