Monitoring

Metrics

Zeebe exports several metrics to facilitate monitoring a cluster. Currently, metrics are exported using Prometheus. You can find documentation about the different metrics here.

Testing

You can easily test metrics locally by using the standard provided docker compose file in combination with the one here, e.g.:

docker-compose --project-directory ./ -f docker-compose.yml -f ../docker/compose/docker-compose.yaml up -d

This will start the usual 3 brokers cluster, as well as a Grafana instance (on port 3000) and a Prometheus instance on port 9090. The Prometheus instance is configured to scrape the brokers every 5 seconds, and pre-assigns them the namespace and pod label as local and broker-*.

Remember that docker-compose does not remove volumes on the down command, so if you are completely done with it you will need to run either docker-compose --project-directory ./ -f docker-compose.yml -f ../docker/compose/docker-compose.yaml down -v or docker volume prune

Grafana

You can find here a pre-built Grafana dashboard to visualize most metrics. This is the dashboard that we use to test and monitor our own Zeebe installations.

NOTE: this dashboard is used for development and can serve as a starting point for your own dashboard, but may not be tailored for your particular use case.

See the Grafana documentation on how to import a dashboard.

Variables

The dashboard at the moment has most visualizations scoped to the following variables: namespace (the k8s namespace), pod (the k8s pod), and partition.

Contributing

If you modify the dashboard, you can modify it using Grafana, then make sure to export as described here, checking Export for sharing externally as you do.

Preview