Monitor PostgreSQL
The DB-Metrics tab shows each node's role, version and health in one overview row, with activity panels below; the Home, Host-Insights and Host-Metrics tabs are described in the common database workflows.
| Overview column | Meaning |
|---|---|
| DB Exporter | Whether the metrics exporter is up |
| DB Status | Whether PostgreSQL is up and reachable |
| DBMode | ReadWrite on the primary, ReadOnly on standbys |
| DBRole | Primary or Standby |
| DBVer | Installed PostgreSQL version, for example 17.5 |
| Datadir | Data directory path |
| Port | Listener port |
| StartTime / UpTime | When PostgreSQL started and how long it has run |
| Archive | WAL archiving status |
| MaxCon | Maximum allowed connections |
Below the overview are panels for Session Status, Session Metrics, Transactions per second, Checkpoints, Replication Delay and WAL activity, per node.
What to watch
-
Replication Delay: how far each standby trails the primary. A delay that keeps growing points to network or resource limits on that standby.
-
Session Status: active sessions per node. A sustained high count suggests connection pooling or tuning.
-
Checkpoints: frequency and duration. Frequent or slow checkpoints hurt write performance.
Validate an observation
Select the correct project, cluster and time range. Confirm the exporter is up before interpreting an empty graph. Compare Primary/Standby and ReadWrite/ReadOnly with current database roles and PostgreSQL cluster-control state when a topology change has just occurred.
Treat growing replication delay, failed WAL archiving, sustained connection saturation and slow checkpoints as signals for investigation. Preserve the time interval and host rather than changing configuration from a single point on a graph.
After deleting a node, stale panels can persist in this release's reported limitations. Check release notes and monitoring troubleshooting.