grafana-dashboards

name	grafana-dashboards
description	Create, import, export, and manage Grafana dashboards for home lab observability.
tools	["homelab_grafanaSnapshot","homelab_prometheusQuery"]
standards-version	1.10.0

Grafana Dashboards

Create, import, and manage Grafana dashboards on a Raspberry Pi home lab. Covers dashboard provisioning, panel configuration, variable templating, data source setup, and JSON export/import workflows.

Trigger

• User asks about creating or editing a Grafana dashboard
• User wants to visualize specific metrics (CPU, memory, disk, network, container stats)
• User mentions "Grafana", "dashboard", "panel", "graph", or "visualization"
• User asks to export or back up a dashboard
• User wants to add a new data source to Grafana
• User asks about Grafana variables or templating

Required Inputs

• SSH connectivity to the Raspberry Pi (validated via homelab_sshTest)
• Grafana running on port 3000

Optional:

• Dashboard UID for export operations
• Specific metrics or PromQL queries for panel creation
• Data source configuration details

Workflow

1. Verify connectivity -- call homelab_sshTest.
2. Check Grafana health -- call homelab_serviceHealth for the Grafana container.
3. For exporting a dashboard:
   • Call homelab_grafanaSnapshot with the dashboard UID
   • Return the JSON configuration for backup or migration
4. For creating a new dashboard:
   • Determine the target metrics using homelab_prometheusQuery to verify data availability
   • Build the dashboard JSON with appropriate panels (graph, stat, gauge, table)
   • Guide the user through provisioning or API import
5. For modifying an existing dashboard:
   • Export the current config via homelab_grafanaSnapshot
   • Apply changes to the JSON structure
   • Re-import via the Grafana API or provisioning directory
6. For data source setup:
   • Guide configuration of Prometheus as the primary data source at http://prometheus:9090
   • Verify connectivity from Grafana to the data source
7. Verify the result -- confirm the dashboard loads correctly and displays data.

Key References

Grafana service:

• Grafana UI: http://pi:3000
• Default credentials: admin/admin (should be changed)
• API endpoint: http://localhost:3000/api/

File locations:

/opt/homelab/docker/monitoring/
  grafana/provisioning/
    dashboards/           # Dashboard provisioning config
    datasources/          # Data source provisioning config
  grafana/dashboards/     # Provisioned dashboard JSON files

Common dashboard UIDs:

• Node Exporter Full: rYdddlPWk
• Docker containers: varies by installation

Useful PromQL for dashboards:

• CPU usage: 100 - (avg by(instance) (irate(node_cpu_seconds_total{mode="idle"}[5m])) * 100)
• Memory usage: (1 - node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes) * 100
• Disk usage: (1 - node_filesystem_avail_bytes{mountpoint="/"} / node_filesystem_size_bytes{mountpoint="/"}) * 100
• Container CPU: rate(container_cpu_usage_seconds_total[5m])

Example Interaction

User: "Export my main Pi dashboard from Grafana"

Assistant actions:

1. Call homelab_sshTest
2. Call homelab_grafanaSnapshot with the dashboard UID
3. Return the JSON config

Response:

Here is the full JSON export for dashboard rYdddlPWk (Node Exporter Full). It contains 24 panels across 6 rows covering CPU, memory, disk, network, and system load. You can re-import this via the Grafana UI at Settings > JSON Model or place it in the provisioning directory.

User: "Create a dashboard showing my Pi's temperature over time"

Assistant actions:

1. Verify data exists: call homelab_prometheusQuery with node_thermal_zone_temp
2. Build a dashboard JSON with a time-series panel
3. Guide the user through import

MCP Usage

Tool	Purpose
homelab_sshTest	Validate connectivity before operations
homelab_grafanaSnapshot	Export dashboard JSON by UID
homelab_prometheusQuery	Verify metric availability for panels
homelab_serviceHealth	Check Grafana container health
homelab_serviceLogs	Debug Grafana issues
homelab_serviceRestart	Restart Grafana after provisioning changes

Common Pitfalls

• Provisioned dashboards are read-only -- dashboards loaded from the provisioning directory cannot be edited in the Grafana UI. Either edit the JSON file and restart Grafana, or save as a new (non-provisioned) dashboard.
• Dashboard UID vs ID -- the API uses UIDs (alphanumeric strings), not numeric IDs. Always use the UID for export and import operations.
• Data source name mismatch -- if a dashboard references a data source by name (e.g., "Prometheus") but the configured name differs, panels will show "No data". Check data source names match exactly.
• Time range confusion -- Grafana defaults to "Last 6 hours". If data appears missing, check the time range selector before investigating Prometheus.
• Variable refresh -- dashboard variables that depend on label values need their refresh setting configured to "On time range change" or "On dashboard load" to stay current.
• Memory on Pi -- complex dashboards with many panels and short refresh intervals can consume significant memory. Keep panel count reasonable and use 30s+ refresh.

See Also

• service-monitoring -- broader monitoring stack overview
• alerting-rules -- Prometheus alerting tied to dashboard metrics
• pi-system-management -- hardware metrics feeding dashboards