Skip to content

Dashboards

The monitoring dashboards provide a hierarchical view of PostgreSQL database health and performance. Users navigate through five levels of detail, from a fleet-wide estate overview down to individual database objects.

Dashboard Hierarchy

The dashboard system organizes metrics into five levels that progress from broad to specific.

  • The estate dashboard shows fleet-wide health across all monitored servers.
  • The cluster dashboard focuses on replication topology and comparative metrics across cluster members.
  • The server dashboard displays system resources and PostgreSQL performance for a single server.
  • The database dashboard presents table and index leaderboards with vacuum status for one database.
  • The object dashboard provides detailed metrics for a specific table, index, or query.

Users navigate between dashboard levels by selecting items in the cluster navigator or by clicking drillable elements within each dashboard. The cluster navigator tree reflects the estate, cluster, server, and database hierarchy.

Time Range Selector

The time range selector controls the time window for all charts in the monitoring section. The selector appears as a toggle button group with the following options:

  • 1h displays the last one hour of data.
  • 6h displays the last six hours of data.
  • 24h displays the last twenty-four hours of data.
  • 7d displays the last seven days of data.
  • 30d displays the last thirty days of data.

The selected time range persists across dashboard navigation. All time-series charts and KPI sparklines update when users change the time range.

Event Timeline

The event timeline displays notable events across the selected servers. The timeline appears above the performance summary tiles in the monitoring section.

The event timeline tracks the following event types:

  • Configuration changes to PostgreSQL settings.
  • Alert activations and resolutions.
  • Server restarts and recovery events.
  • Extension installations and upgrades.
  • Other system-level events.

The event timeline refreshes in sync with the cluster navigator refresh cycle. Users can filter events by server and event type.

AI Chart Analysis

The AI chart analysis feature provides LLM-powered insights for any chart or KPI tile in the monitoring dashboards. The analysis examines data trends, identifies anomalies, and generates actionable recommendations.

Triggering an Analysis

Charts, KPI tiles, leaderboards, and the vacuum status section each display a brain icon. Clicking the icon opens an analysis dialog and starts the LLM analysis.

The analysis follows these steps:

  1. The system checks for a cached analysis result.
  2. The system fetches server context from the connection.
  3. The system fetches timeline events for the time range.
  4. The system serializes the chart data and sends the data to the LLM.
  5. The LLM produces a structured analysis report.

The dialog displays a loading skeleton while the analysis runs. The final report renders as formatted markdown.

Analysis Reports

Each chart analysis report contains a structured assessment of the metric data:

  • The summary section describes the current state of the metric and its significance.
  • The trends and patterns section identifies notable changes, spikes, or anomalies in the data.
  • The recommendations section suggests specific actions to address any issues found.

Timeline Event Correlation

The analysis includes timeline events from the chart's time range to identify correlations between metric changes and system events. The LLM considers the following event types:

  • Configuration changes to PostgreSQL settings.
  • Alert activations and resolutions.
  • Server restarts and recovery events.
  • Extension installations and upgrades.
  • Blackout periods and maintenance windows.

Running SQL Queries

SQL code blocks in analysis reports include a play button in the upper right corner. The run button executes the query against the chart's associated database server. Results appear inline below the code block.

Write statements such as ALTER SYSTEM display a confirmation dialog before executing. Read-only queries execute immediately.

Caching

The system caches chart analysis results on the client side to avoid redundant LLM calls.

  • An amber brain icon indicates that a cached analysis exists for the chart.
  • The cache uses stable identifiers as the cache key; these include the metric description, connection, database, and time range.
  • The cache expires after 30 minutes.
  • Clicking an amber brain icon opens the cached report instantly.

Downloading Reports

The dialog footer includes a Download button that saves the analysis as a markdown file. The downloaded file includes the chart details, the full analysis report, and a generation timestamp.