Get Started with Docker Compose

Use the Docker orchestrator to explore four functions, events, jobs and data in the SlimFaas dashboard. This demonstration uses one SlimFaas node and persistent Docker volumes.

Before you start

Install Git and Docker with Compose v2. Start the engine and allow it to pull .NET base images and build the sample applications. Local .NET and Node installations are not required for container builds.

docker version
docker compose version
git clone https://github.com/SlimPlanet/SlimFaas.git
cd SlimFaas

Keep ports 30021 and 5000 free. The latter is the optional direct host access to fibonacci1; SlimFaas always uses the container's internal port 5000. If host port 5000 is already occupied, choose another before running Compose:

export FIBONACCI_HOST_PORT=5001

In PowerShell, use $env:FIBONACCI_HOST_PORT = '5001'. Stop a native-local demo first because its node ports overlap. SlimFaas manages containers through the mounted Docker socket; this demo is intended for a local development engine.

Build and start

The tutorial overlay extends the repository's Compose example with the remaining Fibonacci functions and correct callback addresses. Build the job image first, then start only the core services. The first command builds FibonacciBatch, the program executed by jobs created through the SlimFaas API. These job containers are created on demand, so docker compose up does not build their image. The tag slimfaas-tour-batch:local must match Configurations.fibonacci.Image in the overlay. The final . supplies the repository root as the build context because FibonacciBatch also references the .NET client under client/dotnet/SlimFaasClient.

docker build -f samples/FibonacciBatch/Dockerfile -t slimfaas-tour-batch:local .
docker compose -f docker-compose.yml -f demo/docker-compose.get-started.yml up -d --build slimfaas fibonacci1 fibonacci2 fibonacci3 fibonacci4
docker compose -f docker-compose.yml -f demo/docker-compose.get-started.yml logs -f slimfaas

The explicit service list leaves Kafka, its connector and Jaeger for their own guides. The overlay enables the data APIs for the tour and configures the fibonacci job to use the image built above. Its fibonacci1 dependency also provides a network for the job.

The overlay replaces the complete jobs configuration, including the required default resource requests and limits (400m CPU and 400Mi memory for this demo). Keep those defaults when adapting the job configuration.

The Fibonacci image declares its internal HTTP port with EXPOSE 5000, allowing SlimFaas to discover the port for every function and managed replica. This does not publish a host port. Its bundled health probe checks HTTP 200 from /health using Bash already present in the runtime image.

Prometheus labels configure scraping for both functions and SlimFaas itself. The runtime is scraped on its internal port 30021, where the queue metrics used by the tour's autoscaling rule are exposed. Keep these labels when adapting the demo so the metrics store and PromQL diagnostics can receive samples.

Leave the services running; Ctrl+C exits the log viewer. SlimFaas can replace the original function containers with managed replicas as it scales, so use the dashboard for the complete function inventory.

Open the dashboard

Open http://127.0.0.1:30021/. In a terminal at the repository root:

export BASE_URL=http://127.0.0.1:30021
curl -i "$BASE_URL/ready"
curl -fsS "$BASE_URL/status-functions"
curl -fsS "$BASE_URL/function/fibonacci1/hello/compose"

Expect 200 READY, the function list and a greeting. Retry readiness while the first startup is in progress. The table shows functions waking and returning to zero after inactivity; the network map shows the route taken by each request.

Continue to the Guided Tour. Select Compose in Bruno.

Docker socket and Podman

By default the example mounts /var/run/docker.sock. For a different socket, set DOCKER_SOCKET_PATH to the host socket path before starting Compose; SlimFaas connects to its mounted path through DOCKER_HOST inside the container.

The repository includes Podman setup helpers. Start your Podman machine and select its connection with podman system connection default <connection> if needed. The helpers use that connection, including machines with custom names, and do not change socket permissions. On macOS:

podman build -f samples/FibonacciBatch/Dockerfile -t slimfaas-tour-batch:local .
./run-podman-compose.sh -f docker-compose.yml -f demo/docker-compose.get-started.yml up -d --build slimfaas fibonacci1 fibonacci2 fibonacci3 fibonacci4

The batch image must be built on the same engine as the Compose services. The helpers mount /run/docker.sock from the VM into SlimFaas; set DOCKER_SOCKET_PATH if your VM uses a different socket path. See the helper's output for socket configuration. On Windows, use the PowerShell helper run-podman-compose.ps1 with the same Compose arguments. The tour's multiline cURL commands use Bash; Bruno provides the same requests on Windows.

Troubleshooting

Symptom	What to check
Cannot connect to Docker	Start Docker/Podman and inspect `docker context show` and the socket path.
Port `5000` is occupied or Podman reports `proxy already running`	Select a free `FIBONACCI_HOST_PORT` and recreate `fibonacci1`; keep the internal port at `5000`.
Function calls time out	Read SlimFaas logs, inspect container health and confirm all tutorial services were built.
Function stays unhealthy with a missing health-check executable	Rebuild the Fibonacci image and recreate services with the current Compose files. The bundled Bash probe checks for HTTP 200 from `/health` without installing extra packages.
Callback never completes	Confirm `SlimFaas__BaseUrl` is `http://slimfaas:30021` inside the function container.
Job cannot start	Build `slimfaas-tour-batch:local` on the same engine SlimFaas uses.
Data API returns 404	Start with both Compose files; the overlay enables public data access for this demo.
A function disappears from Compose's list	SlimFaas creates its own replicas. Inspect the dashboard and `docker ps -a`.

For Kafka, use the Kafka guide; for exported traces, see OpenTelemetry.

Stop and reset

Stop and remove the Compose services:

docker compose -f docker-compose.yml -f demo/docker-compose.get-started.yml down

With Podman, use the same helper for cleanup:

./run-podman-compose.sh -f docker-compose.yml -f demo/docker-compose.get-started.yml down

SlimFaas-created replicas and templates can survive Compose cleanup. Before removing any remaining containers, inspect only the demonstrated function labels and names with docker ps -a; remove the specific demo containers you identify. Do not use a global container prune.

Named data and backup volumes are retained. To reset this dedicated demo, use the same down command with --volumes after removing its remaining managed containers. This deletes the demo's stored data.