mlflow-full/README.md

mlflow-full

MLFlow docker container with support for Postgres. Implements extra environment variables for convenience.

Running the Container

Basic Usage

# Run with default settings
docker run -p 5000:5000 gitea.gt-proj.com/brian/mlflow-full:latest

With PostgreSQL Backend

# Using environment variables
docker run -p 5000:5000 \
  -e MLFLOW_BACKEND_STORE_URI="postgresql://user:password@host:5432/mlflow" \
  -e MLFLOW_DEFAULT_ARTIFACT_ROOT="s3://my-bucket/mlflow-artifacts" \
  gitea.gt-proj.com/brian/mlflow-full:latest

Complete Example with Docker Compose

version: '3.8'

services:
  postgres:
    image: postgres:15
    environment:
      POSTGRES_DB: mlflow
      POSTGRES_USER: mlflow
      POSTGRES_PASSWORD: mlflow_password
    volumes:
      - postgres_data:/var/lib/postgresql/data
    ports:
      - "5432:5432"

  mlflow:
    image: gitea.gt-proj.com/brian/mlflow-full:latest
    ports:
      - "5000:5000"
    environment:
      MLFLOW_BACKEND_STORE_URI: postgresql://mlflow:mlflow_password@postgres:5432/mlflow
      MLFLOW_DEFAULT_ARTIFACT_ROOT: /mlflow/artifacts
      MLFLOW_HOST: 0.0.0.0
      MLFLOW_PORT: 5000
    volumes:
      - mlflow_artifacts:/mlflow/artifacts
    depends_on:
      - postgres

volumes:
  postgres_data:
  mlflow_artifacts:

Environment Variables

Variable	Description	Default
MLFLOW_BACKEND_STORE_URI	Database connection string	sqlite:///mlflow.db
MLFLOW_DEFAULT_ARTIFACT_ROOT	Artifact storage location	./mlruns
MLFLOW_HOST	Host to bind the server	0.0.0.0
MLFLOW_PORT	Port to bind the server	5000

Advanced Examples

With S3 Artifact Storage

docker run -p 5000:5000 \
  -e MLFLOW_BACKEND_STORE_URI="postgresql://user:password@host:5432/mlflow" \
  -e MLFLOW_DEFAULT_ARTIFACT_ROOT="s3://my-mlflow-bucket/artifacts" \
  -e AWS_ACCESS_KEY_ID="your-access-key" \
  -e AWS_SECRET_ACCESS_KEY="your-secret-key" \
  -e AWS_DEFAULT_REGION="us-west-2" \
  gitea.gt-proj.com/brian/mlflow-full:latest

With Custom Server Arguments

docker run -p 5000:5000 \
  -e MLFLOW_BACKEND_STORE_URI="postgresql://user:password@host:5432/mlflow" \
  gitea.gt-proj.com/brian/mlflow-full:latest \
  --host 0.0.0.0 \
  --port 5000 \
  --workers 4 \
  --expose-prometheus /metrics

Kubernetes Deployment

apiVersion: apps/v1
kind: Deployment
metadata:
  name: mlflow-server
spec:
  replicas: 1
  selector:
    matchLabels:
      app: mlflow-server
  template:
    metadata:
      labels:
        app: mlflow-server
    spec:
      containers:
      - name: mlflow-server
        image: gitea.gt-proj.com/brian/mlflow-full:latest
        ports:
        - containerPort: 5000
        env:
        - name: MLFLOW_BACKEND_STORE_URI
          value: "postgresql://mlflow:password@postgres-service:5432/mlflow"
        - name: MLFLOW_DEFAULT_ARTIFACT_ROOT
          value: "s3://mlflow-artifacts"
---
apiVersion: v1
kind: Service
metadata:
  name: mlflow-service
spec:
  selector:
    app: mlflow-server
  ports:
  - port: 5000
    targetPort: 5000
  type: LoadBalancer

Features

• 🐘 PostgreSQL Support: Pre-installed psycopg2-binary for PostgreSQL connectivity
• 🏷️ OCI Labels: Fully compliant with OCI image specification
• 🔄 Auto-updates: Daily builds when the base MLflow image is updated
• 📦 Lightweight: Based on official MLflow container

Building Locally

# Clone the repository
git clone https://gitea.gt-proj.com/brian/mlflow-full.git
cd mlflow-full

# Build the image
docker build -t mlflow-full:local .

# Run your local build
docker run -p 5000:5000 mlflow-full:local