Managed PostgreSQL Providers
pgwatch ships a preset per managed PostgreSQL provider that uses only the metrics the platform's low-privilege monitoring user can see. Pick the preset that matches your platform; the gatherer will then keep the dashboards populated and the log quiet.
Provider comparison
| Provider | Preset | Python / OS helpers | pg_monitor role |
Notes |
|---|---|---|---|---|
| Google Cloud SQL for PostgreSQL | gce |
No | Yes | OS metrics available via Grafana Stackdriver data source. Set track_io_timing and track_functions in the Cloud Console Flags section. |
| Amazon RDS for PostgreSQL | rds |
No | Yes | OS metrics available via Grafana CloudWatch data source. |
| Amazon Aurora PostgreSQL-compatible | rds |
No | Yes | Some PostgreSQL metrics are missing compared to standard RDS. |
| Azure Database for PostgreSQL | azure |
No | Yes | OS metrics available via Grafana Azure Monitor data source. Some file-access functions (e.g. for wal_size) are whitelisted. pg_stat_statements is not activated by default — enable it manually or via the API. |
| Aiven for PostgreSQL | provider-specific setup | No | Yes | See the Aiven developer documentation. |
Common caveats across providers
- PL/Python and other untrusted procedural languages are disabled on every managed platform listed above — Python-based OS helper metrics will fail. Use the platform's OS-metrics pipeline instead.
- The
DB overviewdashboard may show errors for unprivileged users; switch to theDB overview Unprivilegeddashboard when running with a non-superuser role. - Set
track_io_timingandtrack_functionswherever the platform exposes them — many metrics rely on these being on.