HELP!!1! The database is slow!!
summary
A practical, checklist-style method for emergency DB slowdowns that starts with precise questions and moves to evidence: use PG watch dashboards and OpenTelemetry traces to find whether 3rd‑party code or a 264 ms trace is to blame, inspect Postgres logs and pg_stat_activity, run EXPLAIN/ANALYZE, and check connection pooling and idle‑in‑transaction sessions — killing one stuck transaction often restores service. Look for 33% IO wait, 90% CPU or exhausted connections, fix with vacuum/PG Squeeze, raise statistics_target, add IOPS or resize instances. vacuum full and OpenTelemetry traces
AI-written summary · about
chapters (8)
- 0:00Introduction and welcome
- 3:34Understanding slow database reports
- 9:19Is it really the database?
- 18:49Fixing single query issues
- 26:39When everything is slow
- 34:09Low-hanging fruit and sudden drops
- 41:17Gradual performance degradation
- 47:17Q&A and closing
related talks
47mIdentifying Slow Queries and Fixing Them!
26mTroubleshooting High CPU Utilization in PostgreSQL Databases
41mOptimizing your app by understanding your PostgreSQL database
19mKeeping High-Traffic Systems Running When Things Go Wrong
27mPostgres Performance in 15 Minutes
23mBest Practices for Tuning Slow Postgres Queries
45mDebugging Databases for Developers
57mPostgreSQL When It's Not Your Job