Dashboards (Grafana)¶
Grafana provides real-time visualizations of the BBC News ETL pipeline metrics collected by Prometheus. Dashboards allow developers and operators to quickly assess pipeline health, throughput, and failures.
1. Overview¶
- Grafana connects to Prometheus as the data source.
- Provides custom dashboards for producers, consumers, RabbitMQ, and overall pipeline performance.
- Dashboards are pre-configured via JSON files or Helm charts.
- Supports alerts and notifications (Slack, email, etc.) for critical conditions.
2. Example Dashboards¶
| Dashboard | Description |
|---|---|
| Producer Overview | Shows articles scraped, work queue length, scrape failures, and retries by section. |
| Consumer / ETL Dashboard | Displays messages consumed, ETL success/failure counts, retry attempts, and DLQ messages. |
| RabbitMQ Queue Status | Queue depth, message rate, DLQ count, consumer count. |
| Pipeline Health | Combined metrics for producers, consumers, and RabbitMQ, including scaling events (KEDA). |
3. Integration¶
- Grafana is included as a Docker container for local development or deployed via Helm chart in Kubernetes.
- Dashboards can be imported/exported using JSON.
- Use templating variables to switch between sections, queues, or environments dynamically.
4. Best Practices¶
- Keep critical metrics at the top for quick visibility.
- Configure alerts for DLQ accumulation, work queue backlog, or high ETL failure rate.
- Use annotations to mark deployments or incidents on dashboards.
- Combine logs and metrics for root-cause analysis.