Building¶

Go (Controller + Workers)¶

# Build both binaries
docker run --rm -v "$(pwd)/controller":/src -w /src golang:1.25-alpine sh -c \
  "CGO_ENABLED=0 go build -o bin/controller ./cmd/controller/ && \
   CGO_ENABLED=0 go build -o bin/worker ./cmd/worker/"

Output: controller/bin/controller and controller/bin/worker (static binaries).

Python (ML Service)¶

# Build Docker image
docker build -t staffops-anomaly-ml ./ml

Full Stack (docker-compose)¶

# Build + start everything
./scripts/start.sh

# Or manually:
docker compose -f scripts/docker-compose.yaml up --build -d

Docker Images¶

Image	Base	Size
Controller	`alpine:3.20`	~15MB
Worker	`alpine:3.20`	~15MB
ML Service	`python:3.11-slim`	~400MB

Controller/Worker Dockerfile¶

FROM golang:1.25-alpine AS builder
WORKDIR /src
COPY go.mod go.sum ./
RUN go mod download
COPY . .
RUN CGO_ENABLED=0 go build -o /bin/controller ./cmd/controller/
RUN CGO_ENABLED=0 go build -o /bin/worker ./cmd/worker/

FROM alpine:3.20
COPY --from=builder /bin/controller /bin/worker /usr/local/bin/
ENTRYPOINT ["controller"]

ML Dockerfile¶

FROM python:3.11-slim
WORKDIR /app
COPY pyproject.toml .
RUN pip install --no-cache-dir -e .
COPY . .
CMD ["python", "-m", "server.main"]

Multi-Architecture¶

For production images (amd64 + arm64):

docker buildx build --platform linux/amd64,linux/arm64 \
  -t registry/staffops-anomaly-controller:v0.11.0 \
  ./controller

Local development

Local builds are single-arch (your machine's architecture). Multi-arch is only needed for cluster deployment.