values.yaml

Overview

The values.yaml file is a Helm chart values configuration file used to define customizable parameters for deploying a containerized application stack, likely related to a RAG (Retrieval-Augmented Generation) flow platform. This file specifies environment variables, container image details, resource configurations, storage options, service types, and ingress settings for multiple components such as RAGFlow, document engines (Elasticsearch, OpenSearch, Infinity), MinIO, MySQL, and Redis.

This configuration file acts as the primary source of deployment customization, enabling users or operators to control how the application and its dependent services are deployed in a Kubernetes environment. It supports setting passwords, image versions, resource requests, and service exposure types, among others, facilitating both local development and production-grade deployments.

Detailed Explanation

File Structure and Key Sections

The file is structured in YAML format with the following major sections:

• Global Configuration

• Environment Variables (env)

• Component-Specific Configuration

• Ingress Configuration

1. Global Configuration

imagePullSecrets: []

Purpose:
Defines global image pull secrets for pulling container images from private registries. This is a list of references to Kubernetes secrets.

Usage:

• Empty by default; can be populated with secrets to authenticate image pulls.

2. Environment Variables (env)

This section defines environment variables that configure various services and their behavior.

Key Variables:

3. Component-Specific Configuration

Each service/component has its own nested configuration specifying container images, storage, deployment, and service details.

3.1 ragflow

ragflow:
  image:
    repository: infiniflow/ragflow
    tag: v0.20.5-slim
    pullPolicy: IfNotPresent
    pullSecrets: []
  service_conf: 
  llm_factories:
  deployment:
    strategy:
    resources:
  service:
    type: ClusterIP
  api:
    service:
      enabled: true
      type: ClusterIP

• Purpose: Configuration for the RAGFlow application container.

• Image: Container image and tag.

• service_conf: Optional overrides written into local.service_conf.yaml inside the container.

• llm_factories: Optional JSON/YAML overrides for llm_factories.json inside the container, describing LLM factories and capabilities.

• deployment: Kubernetes deployment strategy and resource requests/limits.

• service: Kubernetes service type; by default ClusterIP.

• api.service: API service enablement and service type.

Usage example:

To expose the RAGFlow API externally, set ragflow.api.service.enabled: true and change type to LoadBalancer.

3.2 infinity

infinity:
  image:
    repository: infiniflow/infinity
    tag: v0.6.0-dev5
    pullPolicy: IfNotPresent
    pullSecrets: []
  storage:
    className:
    capacity: 5Gi
  deployment:
    strategy:
    resources:
  service:
    type: ClusterIP

• Purpose: Configuration for the Infinity document engine.

• Storage: Persistent volume claim class and capacity (default 5Gi).

• Deployment and service: Kubernetes deployment strategy, resources, and service type.

3.3 elasticsearch

elasticsearch:
  image:
    repository: elasticsearch
    tag: "8.11.3"
    pullPolicy: IfNotPresent
    pullSecrets: []
  initContainers:
    alpine:
      repository: alpine
      tag: latest
      pullPolicy: IfNotPresent
    busybox:
      repository: busybox
      tag: latest
      pullPolicy: IfNotPresent
  storage:
    className:
    capacity: 20Gi
  deployment:
    strategy:
    resources:
      requests:
        cpu: "4"
        memory: "16Gi"
  service:
    type: ClusterIP

• Purpose: Configurations for Elasticsearch, including image versions and resource requests.

• Init containers: Alpine and Busybox images for initialization tasks.

• Storage: 20Gi persistent volume capacity by default.

• Resources: Requests for CPU and memory optimized for Elasticsearch workload.

3.4 opensearch

Similar to Elasticsearch, but for OpenSearch service with version 2.19.1.

3.5 minio

Configuration for MinIO object storage with a default 5Gi storage.

3.6 mysql

MySQL database configuration with 5Gi storage and MySQL image version 8.0.39.

3.7 redis

Redis cache configuration with 5Gi storage, persistence enabled, and retention policy commented out.

4. Ingress Configuration

ingress:
  enabled: false
  className: ""
  annotations: {}
  hosts:
    - host: chart-example.local
      paths:
        - path: /
          pathType: ImplementationSpecific
  tls: []

• Purpose: Setup for Kubernetes ingress resources to expose web services externally.

• Defaults: Disabled by default.

• Annotations: Allows specifying ingress controller annotations (e.g., nginx).

• Hosts: List of hostnames and paths for ingress routing.

• TLS: Configuration for TLS certificates.

Important Implementation Details

• Password Complexity: The file enforces complex passwords for OpenSearch, requiring uppercase, lowercase, digits, and special characters.

• Batch Processing Parameters: DOC_BULK_SIZE and EMBEDDING_BATCH_SIZE control batch sizes for document chunk parsing and embedding vectorization respectively, influencing performance and resource use.

• Image Pull Policies: Most images use IfNotPresent to avoid unnecessary pulls.

• Service Exposure: Defaults to ClusterIP to restrict access within the cluster; can be changed to LoadBalancer for external exposure.

• Storage Classes: Storage class names are left empty by default, allowing customization depending on the Kubernetes environment.

• Init Containers: Used in Elasticsearch and OpenSearch to run lightweight containers before main containers start, potentially for setup or validation.

Interaction With Other System Components

• This file is used by Helm during the deployment process to render Kubernetes manifests with appropriate values.

• It configures multiple interdependent services:

  • RAGFlow as the main application.

  • Document Engines (Elasticsearch, Infinity, OpenSearch) for indexing and searching documents.

  • MinIO for object storage.

  • MySQL for relational data persistence.

  • Redis for caching and fast data retrieval.

• The environment variables defined here configure runtime behavior of containers.

• The ingress configuration allows external access to the deployed services.

• Storage configurations ensure data persistence across pod restarts.

• Passwords and secrets here secure the services and are typically overridden in production environments or managed via Kubernetes Secrets.

Usage Example

To deploy with Elasticsearch as the document engine and expose the RAGFlow service externally with LoadBalancer:

env:
  DOC_ENGINE: elasticsearch

ragflow:
  service:
    type: LoadBalancer

elasticsearch:
  service:
    type: LoadBalancer

Visual Diagram: Flowchart of Main Configuration Sections and Relationships

flowchart TD
    A[values.yaml] --> B[Global Config]
    A --> C[Environment Variables (env)]
    A --> D[RAGFlow Config]
    A --> E[Document Engines Config]
    A --> F[Storage Services Config]
    A --> G[Ingress Config]

    E --> E1[Elasticsearch]
    E --> E2[OpenSearch]
    E --> E3[Infinity]

    F --> F1[MinIO]
    F --> F2[MySQL]
    F --> F3[Redis]

    D --> D1[Image & Tag]
    D --> D2[Service Type]
    D --> D3[Deployment Resources]

    G --> G1[Enabled]
    G --> G2[Hosts & Paths]
    G --> G3[TLS Settings]

Summary

The values.yaml file is a comprehensive Helm chart configuration for deploying a RAGFlow-based application stack on Kubernetes. It centralizes configuration for multiple services including document engines, storage, database, caching, and ingress, enabling flexible customization of deployment parameters like image versions, passwords, resource requests, and service exposure. It plays a critical role in defining the behavior and security of the deployed application components.