Docker Uv Image Builds

Docker Image Builds for Python Services

Image naming

Images follow the pattern: mitodl/

The service name should match the application or code location name used elsewhere in configuration (Helm values, Pulumi stacks, Concourse pipelines).

Image tags

Tag images with the git short ref (7-character SHA):

GIT_TAG=$(git rev-parse --short HEAD)
docker build -t mitodl/${SERVICE_NAME}:${GIT_TAG} .

Do not use latest as a production tag.

Python environment inside the image

Use a relocatable virtual environment so the venv works after Docker layer assembly:

WORKDIR /app
COPY pyproject.toml uv.lock ./
RUN uv venv --relocatable /app/.venv && \
    uv sync --frozen --no-dev

Shared libraries

Install shared internal libraries (e.g. ol-orchestrate-lib) as build-time dependencies — do not mount them as Docker volumes at runtime. Add them to pyproject.toml and let uv sync install them during the image build.

Build context & .dockerignore

Exclude development artifacts:

.venv/
__pycache__/
*.pyc
.git/

Concourse CI integration

Concourse pipelines use paths: filters on git resources to trigger image rebuilds only when relevant files change. When adding a new service, add its path to the corresponding pipeline's git resource paths list.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: mitodl
• Source: mitodl/agent-kit
• License: BSD-3-Clause

Install and usage instructions live in the source repository linked above.