pgAssistant Auto-Tune is now available. It turns selected pgAssistant recommendations into controlled actions, for a single PostgreSQL database or an entire fleet.
pgAssistant advises, Collector stores recommendation snapshots, Grafana makes fleet priorities visible, and Auto-Tune executes the actions you authorize.
Choose, preview, execute
Define a YAML policy to select recommendations by team, advisor, action type, schema, table, or confidence. Apply a common policy across the fleet, with overrides for individual databases.
Start in dry-run mode to inspect the proposed SQL without connecting to target databases or executing changes. When ready, switch to execution mode and submit a new run. Starting the services alone does not trigger tuning.
Supported actions include:
- Indexes: concurrent creation and removal through supported advisor recommendations.
- Maintenance:
ANALYZEandVACUUM. - Autovacuum configuration: supported changes that do not require a restart.
The Parameter Advisor does not yet have an executor. All profiles in the default policy are initially disabled.
Keep execution visible
The API retains policies, source snapshots, SQL, step results, and durations. Failed or uncertain attempts require review before retrying; multi-step scripts have no automatic rollback.
To get started, you need Docker Compose and a pgAssistant Collector collecting the executive_plan job.