Skip to main content
View rawEdit

07. Autovacuum and estimated bloat

Monitor autovacuum effectiveness and track bloat accumulation across your database.

note

This dashboard is currently under development. In the live demo, it shows limited data. The underlying metrics are being collected as part of health check data, and the dashboard will be expanded in a future release.

When to use

  • Checkup reports autovacuum settings need attention (F001)
  • You suspect autovacuum is not keeping up with dead tuple generation
  • Investigating why a table keeps growing despite stable row counts
  • Tuning autovacuum settings for high-write tables

Key panels (current and planned)

  • Vacuum timeline — when vacuums ran and on which tables
  • Dead tuple accumulation — rate of dead tuple generation vs cleanup (planned)
  • Autovacuum worker utilization — how many workers are active vs available (planned)
  • Bloat estimates — per-table and per-index bloat percentages (planned)

What good looks like

  • Autovacuum runs regularly on large, frequently-updated tables
  • Dead tuple ratio stays low (< 10% of live tuples for most tables)
  • Autovacuum workers are not saturated
  • Bloat percentages are within acceptable ranges (varies by table size)

What to investigate

SignalNext step
Dead tuples accumulatingCheck if autovacuum is running; look for long-running transactions holding back cleanup
Autovacuum workers maxed outIncrease autovacuum_max_workers or tune per-table settings
High bloat on specific tableGo to Single table analysis for detailed view
Vacuum never runs on a tableCheck if autovacuum_enabled is set to off for that table
  • F001 — autovacuum current settings (CLI)
  • A007 — altered settings (CLI)
  • F004 — heap bloat estimated (monitoring stack)
  • F005 — index bloat estimated (monitoring stack)