Skip to main content
Version: 6.2

Database Observability: Instance State

The Database Observability: Instance State dashboard is intended for initial diagnostics of a selected PostgreSQL instance after an alert on the fleet overview dashboard or in an incident card.

Who it is for: database administrators, on-call engineers, and performance specialists.

Data sources: db_metrics*, db_activity*, db_meta*.

Main Sections​

1. Status Summary​

Shows key metrics for the selected instance:

  • uptime, TPS, and proportion of occupied connections
  • cache hits and rollback rate
  • availability, transactions, and rollbacks over time

Instance status summary

2. Databases​

Shows the selected instance's databases, their sizes, connections, TPS, temporary files, deadlocks, and cache hits.

Instance databases

3. Sessions and Waits​

Shows session status and wait characteristics:

  • distribution of sessions by state and wait type
  • wait duration
  • number of checkpoints

Sessions and waits

4. Input/Output​

Shows which processes generate disk load and how input/output wait time changes.

Disk load

PanelDestination
Uptime metricDatabase Observability: Replication and HA
Rollback Rate metricDatabase Observability: Locks and Transactions
Instance Databases tableDatabase Observability: Queries