Fred

Backend Deployment Guide

Deployment Guide

Fred can be deployed in multiple ways depending on your needs:

  • For local development on your laptop
  • With Docker Compose for easier multi-container management
  • On Kubernetes using Helm charts, recommended for production or advanced testing

No matter the deployment method, Fred requires two main configuration sources:

  1. config/configuration.yaml — enables or disables Fred’s internal services.
  2. .env file — provides external connectors and credentials (OpenSearch, PostgreSQL/pgVector, Minio, Azure, Keycloak, etc.).

When working in development mode, you should maintain your own .env file locally and make sure it’s ignored by Git (included in .gitignore) to avoid leaking secrets.

In Kubernetes environments, the .env file is converted into Kubernetes Secrets and mounted at runtime for security.

Local Development

This method is ideal for developers working directly on the Fred codebase. After cloning the repository:

  1. Copy the .env.template to create your local .env:

    cp config/.env.template config/.env

  2. Edit config/.env to fill in your actual credentials.

  3. Launch the backend locally:

    cd backend
make run

  4. The frontend can be launched separately (if applicable).

Tip: your .env is never committed as specified in the repository .gitignore.

Docker Compose

This method is great for testing multi-container setups locally without Kubernetes.

  1. Copy the .env.template and configure it:
   cp config/.env.template config/.env

Build and run the Docker container using the Makefile:

make docker-build
make docker-run

Docker will orchestrate Fred inside a container, and you can test integrations with services like Postgres, Minio, OpenSearch, etc.

Check the deployfolder. A docker compose file is available that will start the various components required by Fred.

Kubernetes with Helm

Helm charts are provided for easier deployment to Kubernetes.

Use this method for production environments or advanced integrations where Fred relies on external services like vector databases, S3-compatible storage, etc.

  1. Prepare your values and secrets:

    • Use the .env.template to prepare your secrets.
    • Encrypt your .env as a Kubernetes Secret.

  2. Use kastctl to provision that faile as an encrypted Kubenetes secret.

  3. Fred will be deployed along with its dependencies (if defined in your Helm chart).

Configuration Overview

backend/resource/configuration.yaml

This file defines which internal services of Fred are active. Particularly important is the ài`section where you can define your agents team.

database:
  type: csv
  csv_files:
    # Can be absolute paths or relative paths to the main
    energy_mix: './services/cluster_consumption/data/simulated_energy_mix.csv'
    carbon_footprint: './services/cluster_consumption/data/simulated_cluster_consumption_gco2_sep_to_feb.csv'
    energy_footprint: './services/cluster_consumption/data/simulated_cluster_consumption_wh_sep_to_feb.csv'
    financial_footprint: './services/cluster_consumption/data/simulated_cluster_consumption_usd_sep_to_feb.csv'

kubernetes:
  kube_config: '~/.kube/config'
  aws_config: '~/.aws/config' # Optional, needed for aws EKS clusters.
  # Timeout settings for the client
  timeout:
    connect: 5  # Time to wait for a connection in seconds
    read: 15    # Time to wait for a response in seconds

ai:
  # Timeout settings for the client
  timeout:
    connect: 5  # Time to wait for a connection in seconds
    read: 15    # Time to wait for a response in seconds
  default_model:
    #provider: "ollama"
    #model: "llama2"
    provider: "openai"
    name: "gpt-4o"
    temperature: 0
  leader:
      enabled: true
      model: {}
  agents:
    GeneralistExpert:
      enabled: true
      model: {}
    DocumentsExpert:
      enabled: true
      categories:
        - "eco-conception"
      settings:
        document_directory: "./resources/knowledge/imported"
        chunk_size: 512
        chunk_overlap: 64
      model: {}
    TechnicalKubernetesExpert:
      enabled: true
      categories:
        - "kubernetes"
        - "namespaces"
        - "workloads"
        - "architecture"
        - "security"
        - "networking"
        - "storage"
        - "configuration"
        - "scaling"
        - "deployment"
      model: {}
    MonitoringExpert:
      enabled: true
      categories:
        - "monitoring"
        - "observability"
        - "logging"
        - "metrics"
        - "electricity"
        - "resources"
        - "consumption"
      model: {}

# Where to save fred produced resources like Essentials or Score
# and external resources like Kubernetes Workload descriptions
dao:
  type: "file"  # Currently the only one supported
  base_path: "~/.fred/fred-backend-cache"
  max_cached_delay_seconds: 10  # Cache delay in seconds. Use 0 for no cache or a negative value for limitless cache.

# Enable or disable the security layer
security:
  enabled: false
  keycloak_url: "http://fred-keycloak:8080/realms/fred"

Adjust according to your needs.

.env file

This file provides connection details to external services.

Fred requires access to:

  • Azure for AI services
  • Keycloak for authentication
  • OpenSearch for metadata, vector search, and chat memory
  • Minio for S3-compatible storage
  • Postgres/pgVector if applicable

Here is the .env.template provided in the repository:

# AZURE
AZURE_TENANT_ID=""
AZURE_CLIENT_ID=""
AZURE_CLIENT_SECRET=""
AZURE_CLIENT_SCOPE=""
AZURE_APIM_KEY=""
AZURE_API_VERSION=""
AZURE_DEPLOYMENT_LLM=""
AZURE_DEPLOYMENT_EMBEDDING=""

# KEYCLOAK
KEYCLOAK_SERVER_URL="https://localhost:8080"
KEYCLOAK_REALM_NAME="fred"
KEYCLOAK_CLIENT_ID="fred"

# OPENSEARCH
OPENSEARCH_HOST="https://localhost:9200"
OPENSEARCH_USER="admin"
OPENSEARCH_PASSWORD="Azerty123_"
OPENSEARCH_SECURE=true
OPENSEARCH_METADATA_INDEX="bidgpt-dev-metadata"
OPENSEARCH_VECTOR_INDEX="bidgpt-dev-embeddings"
OPENSEARCH_CHAT_MEMORY_INDEX="bidgpt-dev-chat-memory"
OPENSEARCH_VERIFY_CERTS=false

# MINIO
MINIO_ENDPOINT="localhost:9000"
MINIO_ACCESS_KEY="admin"
MINIO_SECRET_KEY="Azerty123_"
MINIO_BUCKET_NAME="bidgpt-dev-content"
MINIO_SECURE=false

Contact Fred’team

Fred is an open source project that serves as basis for production grade application. We maintain CI/CD to deploy Fred instances on GCP, on Azure, or on internal Kubernetes clusters. Get in touch with us.

Next
Getting Started