Skip to main content

Verify Installation

After starting the Docker Compose stack, walk through these checks to confirm everything is working.

1. Check running containers​

docker compose ps

All services should show Up (healthy):

NAME STATUS
postgres Up (healthy)
minio Up (healthy)
temporal Up (healthy)
yardrail-frontend Up (healthy)
yardrail-worker Up
note

The worker doesn't have a health check (it's a pure Temporal task-queue poller with no HTTP surface). Up without (healthy) is expected.

2. Check the frontend​

Open http://localhost:8080 — you should see the Yardrail login page.

3. Check the worker logs​

docker compose logs yardrail-worker --tail=10

A healthy worker logs its startup configuration and begins polling:

yardrail-worker | {"level":"INFO","msg":"yardrail-worker starting",
"temporal":"temporal:7233","executor":"docker","s3_endpoint":"minio:9000",...}

Common startup failures​

SymptomCauseFix
dial tcp temporal:7233: connection refusedTemporal not ready yetWait for Temporal to be healthy, then restart: docker compose restart yardrail-worker
worker: s3 client: ...MinIO not reachableCheck MinIO is healthy: docker compose ps minio
worker: docker client: Cannot connect to the Docker daemonDocker socket not mountedAdd socket mount — see step 4 of the install guide

4. Check Docker socket access​

If you've configured the socket mount, verify the worker can reach the Docker daemon:

docker compose exec yardrail-worker docker version

If this fails, the docker-compose.override.yml socket mount isn't in place. The worker will start and connect to Temporal, but workflow tool execution will fail.

5. Check S3 storage​

Verify the workflow artifacts bucket exists:

docker compose exec minio mc ls local/yardrail-workflows

The bucket is created automatically by the worker on first startup. If it doesn't exist, the worker couldn't reach MinIO — check the logs.

6. Run a test workflow​

Trigger a workflow through the Yardrail UI, API, or the yr CLI:

yr auth login # authenticate against http://localhost:8080
yr template list # confirm templates are visible
yr installation trigger <id> --wait # trigger a run and stream the journal

You can also watch the worker logs to see it pick up the task:

docker compose logs yardrail-worker -f

When a CWL container tool runs, you should see a new Docker container appear briefly:

# In another terminal
docker ps --filter "label=app.kubernetes.io/managed-by=yardrail-worker"
note

The label filter above works for the Kubernetes executor. For the Docker executor, tool containers are created directly via the Docker API without Yardrail-specific labels. Use docker ps and look for containers with the CWL tool's image name.

After the workflow completes, check that outputs landed in S3:

docker compose exec minio mc ls local/yardrail-workflows/workflows/ --recursive | head

Troubleshooting​

Worker logs show "context deadline exceeded" during workflow execution​

The CWL tool container is taking longer than the configured timeout. Check the workflow definition's timeout settings.

Workflow outputs are empty​

Check that the S3 tar-streaming pipeline is working:

  1. Worker should log no errors during CopyToContainer / CopyFromContainer
  2. Verify the MinIO bucket has objects under workflows/{uuid}/out/
  3. Check the tool container's exit code in the worker logs — a non-zero exit may mean the tool failed before producing output

Container images not found​

The Docker executor pulls images on first use. If the image is in a private registry, configure Docker credentials on the host (via docker login or ~/.docker/config.json). The worker uses the host's Docker credential store.