LibreChat Docker Compose Service Breakdown | Generated by AI

Home 2025.09 English 中文 日本語 Español हिंदी Français Deutsch العربية 繁體中文

---

What you pasted is a docker-compose.yaml setup for running LibreChat along with its dependencies. Let me break it down clearly:

---

High-Level View

The setup defines 6 services (containers) that work together:

1. LibreChat API/UI (api)
2. MongoDB (mongodb)
3. Meilisearch (meilisearch)
4. Postgres with pgvector (vectordb)
5. RAG API (rag_api)
6. Volumes for persistent storage

These are all wired together using Docker Compose so that you can start everything with a single docker-compose up.

---

Service Explanations

1. LibreChat API (api)

• Container name: LibreChat
• Image: ghcr.io/danny-avila/librechat-dev:latest (the dev build of LibreChat)
• Ports: Exposes ${PORT} from .env to your host machine.
• Dependencies: Waits for mongodb and rag_api before starting.

• Environment variables:

  • MONGO_URI: Connection string for MongoDB.
  • MEILI_HOST: Points to Meilisearch service.
  • RAG_API_URL: Connects to the local RAG API container.

• Volumes:

  • .env file is mounted inside the container.
  • images, uploads, and logs are persisted locally so you don’t lose them when the container restarts.

👉 This is the main app you interact with (the LibreChat web/API service).

---

2. MongoDB (mongodb)

• Container name: chat-mongodb
• Image: mongo (official MongoDB image).
• Purpose: Stores chat data, user sessions, config, etc.
• Command: Runs mongod --noauth (no authentication).
• Volumes: ./data-node:/data/db so your database persists outside the container.

---

3. Meilisearch (meilisearch)

• Container name: chat-meilisearch
• Image: getmeili/meilisearch:v1.12.3
• Purpose: Provides fast full-text search capabilities for LibreChat.

• Environment variables:

  • MEILI_NO_ANALYTICS=true (privacy).
  • MEILI_MASTER_KEY is pulled from .env.
• Volumes: ./meili_data_v1.12:/meili_data for persistence.

---

4. Vector Database (vectordb)

• Container name: vectordb
• Image: pgvector/pgvector:0.8.0-pg15-trixie (Postgres with pgvector extension).
• Purpose: Stores embeddings for RAG (Retrieval Augmented Generation).

• Environment:

  • POSTGRES_DB, POSTGRES_USER, POSTGRES_PASSWORD.
• Volume: pgdata2:/var/lib/postgresql/data for persistence.

---

5. RAG API (rag_api)

• Container name: rag_api
• Image: ghcr.io/danny-avila/librechat-rag-api-dev-lite:latest
• Purpose: Handles RAG workflows — retrieving documents and augmenting model responses with external knowledge.

• Environment:

  • DB_HOST=vectordb (connects to Postgres vector DB).
  • RAG_PORT defaults to 8000.
• Depends on: vectordb.
• Env file: Reuses your .env.

---

6. Volumes

• pgdata2 → Persistent data for Postgres/pgvector.
• Other volumes like ./meili_data_v1.12, ./data-node, and mounted directories (images, uploads, logs) ensure your data survives restarts.

---

How They Work Together

1. LibreChat (UI/API) is the central service you interact with.
2. It stores chat/session data in MongoDB.
3. It queries Meilisearch for text search features.
4. For RAG (document retrieval), LibreChat calls rag_api.
5. The RAG API fetches vector embeddings from vectordb (Postgres + pgvector).
6. Everything persists on local volumes, so data is not lost when containers restart.

---

✅ In short:

• MongoDB → chat/user data
• Meilisearch → fast keyword search
• Postgres+pgvector → vector embeddings for RAG
• RAG API → retrieval augmentation layer
• LibreChat → the app tying it all together

---

Do you want me to also show you a diagram-style flow (textual) of how requests move through these containers (from you → LibreChat → RAG API → Vector DB)? That might make it even clearer.

---

Back

openai/gpt-5

Donate