Docker Compose Reference¶

WitFoo Analytics provides five Docker Compose files for different deployment topologies. All compose files are located in the docker/ directory.

Compose Files¶

All-in-One (AIO)¶

The most common deployment. Runs all 11 Analytics services on a single node.

cd docker
docker compose -f docker-compose.aio.yml up -d --build

Services Included¶

Startup Order¶

Services start in dependency order enforced by health checks:

cassandra, nats (foundation)
    --> artifact-ingestion, graph-processor, incident-engine (processing)
        --> api, dispatcher (application)
            --> reverse-proxy (routing)
                --> ui (frontend)
                    prometheus --> grafana (observability)

AIO + Conductor¶

Combines all Analytics and Conductor services in a single compose file for single-node deployments with log collection.

cd docker
docker compose -f docker-compose.aio-conductor.yml up -d --build

This adds five Conductor services to the standard AIO deployment. See Conductor Pipeline for details on the Conductor services.

Data Node¶

Deploys a single Cassandra instance configured for multi-node clustering.

NODE_IP=10.0.1.10 CASSANDRA_SEEDS=10.0.1.10,10.0.1.11 \
  docker compose -f docker-compose.data-node.yml up -d --build

Ports exposed: 9042 (CQL), 7000 (inter-node communication).

Processing Node¶

Runs all application services except Cassandra, connecting to external data nodes for storage.

DATA_NODE_HOSTS=10.0.1.10:9042,10.0.1.11:9042 \
  docker compose -f docker-compose.processing-node.yml up -d --build

Conductor Add-on¶

Adds Conductor services to an existing Analytics deployment by joining the external witfoo-analytics-net network.

# Analytics stack must be running first
cd docker
docker compose -f docker-compose-conductor.yml up -d --build

Common Configuration¶

Network¶

All services communicate over the witfoo-analytics-net bridge network. No inter-service traffic is exposed to the host unless explicitly mapped via port bindings.

Logging¶

All services use the same JSON file logging driver:

logging:
  driver: "json-file"
  options:
    max-size: "10m"
    max-file: "3"
    compress: "true"

Volumes¶

Health Checks¶

All application services expose a /health endpoint. Health check configuration:

Cassandra has the longest start period (60-120s) due to JVM warm-up and schema initialization.

Restart Policy¶

All services except initialization containers use restart: unless-stopped. This ensures services restart after crashes but stay stopped if intentionally brought down.

LISTEN_IP¶

The LISTEN_IP environment variable controls which network interface services bind to:

# Bind to all interfaces (default for most services)
LISTEN_IP=0.0.0.0

# Bind to localhost only (default for Prometheus/Grafana)
LISTEN_IP=127.0.0.1

# Bind to a specific interface
LISTEN_IP=10.0.1.10

Build Secrets¶

Go services use BuildKit secrets to access private Git dependencies during build:

secrets:
  git_credentials:
    file: ${HOME}/.git-credentials

Ensure your ~/.git-credentials file contains a valid GitHub personal access token.

Port Summary¶