Skip to main content

Validation

Validating a Cadence installation means confirming that the server is running, the persistence layer is healthy, and workflows can execute end-to-end. The steps below apply to any installation method; specific commands differ by deployment path.

1. Verify server health​

Local binary​

The server prints startup logs to stdout. Look for frontend, history, matching, and worker services binding to their configured ports.

Docker Compose​

Check that all containers are running:

docker compose ps

The cadence container should be Up and healthy. If it is not, check logs:

docker logs -f docker-cadence-1
docker logs -f docker-cassandra-1

Kubernetes / Helm​

Wait for all pods to reach Running and schema setup jobs to complete:

kubectl get pods -l app.kubernetes.io/name=cadence-release -n cadence
kubectl get jobs -l app.kubernetes.io/component=schema-server -n cadence

Schema jobs should show Completed (1/1). If pods are not ready, check schema job logs first, as services depend on schema being in place:

kubectl logs job/cadence-release-schema-server -n cadence

2. Register a test domain and verify​

Domain registration is the first functional check. A successful registration confirms the frontend service is reachable and the persistence layer is operational.

Local CLI (local binary or Docker Compose):

cadence --domain test-domain domain register --retention 1
cadence --domain test-domain domain describe

Kubernetes (no local CLI):

kubectl port-forward svc/cadence-release-frontend 7833:7833 -n cadence
cadence --address localhost:7833 --transport grpc --domain test-domain domain register --retention 1

The domain describe output should show Status: REGISTERED.

3. Run an end-to-end workflow​

The strongest validation is executing a real workflow. The cadence-samples repository provides ready-to-run examples.

Go samples (helloworld):

git clone https://github.com/cadence-workflow/cadence-samples.git
cd cadence-samples

# Start the worker
go run cmd/samples/recipes/helloworld/main.go

# In another terminal, trigger the workflow
cadence --domain test-domain workflow start \
  --tasklist helloworld-tasklist \
  --workflow_type helloworld_workflow \
  --execution_timeout 60

Java and Python samples follow the same pattern: start a worker process, then trigger a workflow via the CLI or the SDK.

After triggering the workflow, verify it completed successfully:

cadence --domain test-domain workflow list

The workflow should appear with a Completed status. The Web UI (default http://localhost:8088) provides a visual confirmation: navigate to the test domain and inspect the workflow execution, history events, and activity results.

4. Validate advanced visibility (optional)​

If Elasticsearch or OpenSearch is deployed, confirm that workflow data is being indexed:

cadence --domain test-domain workflow list --query "WorkflowType = 'helloworld_workflow'"

A successful list query with results confirms that the visibility pipeline (Kafka → Elasticsearch/OpenSearch) is operational. For OpenSearch deployments, you can also check the index directly:

curl http://localhost:9200/cadence-visibility*/_count

5. Validate metrics (production)​

For production deployments, confirm that the metrics pipeline is operational. The default Docker Compose setup includes Prometheus and Grafana. Verify Prometheus is scraping Cadence targets:

  • Prometheus: navigate to http://localhost:9090/targets and confirm Cadence endpoints are UP.
  • Grafana: navigate to http://localhost:3000 and check the pre-configured Cadence dashboards for incoming data.

For Kubernetes deployments using Prometheus Operator or GMP (Google Managed Prometheus), verify that PodMonitor or ServiceMonitor resources are picking up Cadence pods. The Helm chart includes an example PodMonitoring values file for GKE environments.

Cadence emits metrics for all four services (frontend, history, matching, worker). Seeing request-count and latency metrics after running a test workflow confirms end-to-end metrics collection.

6. Operational validation (production)​

For production deployments, Cadence provides two additional validation tools:

  • Canary workflows run periodic health-check workflows that exercise core server functionality. Deploy via docker compose -f docker-compose-canary.yml up or as a separate Kubernetes workload. See the canary README.
  • Bench workflows run load tests against the cluster to validate capacity and performance baselines. Deploy via docker compose -f docker-compose-bench.yml up. See the bench README.

Both tools run as Cadence workers that start workflows against the cluster, providing ongoing confidence that the system is functioning correctly under load.