Dashboarding and Alerting
This page is a quick orientation to the dashboarding and alerting tooling that pairs with pgwatch. For background and trade-offs see Concept: Observability stack. For the concrete alert-setup recipe see How-to: Set up alerting.
Where the dashboards live
pgwatch ships a set of pre-defined Grafana dashboards covering most of the metrics the built-in presets collect. Browse them in the Gallery → Dashboards page or in the grafana/ folder of the repository. The dashboards assume one of these data sources is configured:
pgwatch-metrics— for Postgres / TimescaleDB sinkspgwatch-prometheus— for the Prometheus sink
The default Docker image wires both for you.
Customising dashboards
Almost every deployment ends up tweaking the built-in dashboards (colours, units, panel layout). Two practices keep that work portable across upgrades:
- Use Save as to put customised dashboards in a separate folder rather than overwriting the originals.
- Treat the built-in dashboards as code you can pull and re-import after pgwatch upgrades.
See Concept: Operating in production → Dashboard maintenance for the long-term playbook.
Alerting
Grafana's built-in alerting is the path of least resistance and is fully compatible with pgwatch's stored metrics. The recipe lives at How-to: Set up alerting.