Implementation:Langgenius Dify Docker Compose Up

Knowledge Sources	Dify
Domains	DevOps, Deployment, Docker, Orchestration
Last Updated	2026-02-12 00:00 GMT

Overview

Concrete tool for launching the complete Dify application stack using docker compose up -d, provided by the docker/docker-compose.yaml orchestration file.

Description

The docker compose up -d command reads the docker-compose.yaml file (lines 697-1626) along with the .env file and starts all services in detached mode. The compose file defines the following core service topology:

Service	Image	Role	Port	Key Configuration
api	langgenius/dify-api:1.13.0	Flask + Gunicorn (gevent) REST API	5001 (internal)	MODE=api, shared env anchor
worker	langgenius/dify-api:1.13.0	Celery worker for background tasks	None	MODE=worker, shared env anchor
worker_beat	langgenius/dify-api:1.13.0	Celery Beat periodic scheduler	None	MODE=beat, shared env anchor
web	langgenius/dify-web:1.13.0	Next.js frontend via PM2	Internal only	PM2_INSTANCES=2
nginx	nginx:latest	Reverse proxy	80, 443	Routes /api to api, / to web
db_postgres	postgres:15-alpine	PostgreSQL metadata DB	5432 (internal)	Profile: postgresql
redis	redis:6-alpine	Cache + Celery broker	6379 (internal)	Password-protected
sandbox	langgenius/dify-sandbox:0.2.12	Code execution sandbox	8194 (internal)	SSRF proxy network
plugin_daemon	langgenius/dify-plugin-daemon:0.5.3-local	Plugin lifecycle manager	5002, 5003	DIFY_INNER_API_KEY for auth
ssrf_proxy	ubuntu/squid:latest	Squid SSRF protection proxy	3128 (internal)	Internal network only
init_permissions	busybox:latest	Volume ownership init	None	Runs once, exits

The startup sequence follows the dependency graph:

init_permissions runs and completes (sets UID 1001 on storage)
db_postgres starts and becomes healthy (pg_isready)
redis starts
api, worker, worker_beat start (depend on init, db, redis)
web starts (no dependencies)
nginx starts (depends on api and web)
sandbox, ssrf_proxy, plugin_daemon start in parallel

Usage

Run this command from the docker/ directory after configuring the .env file. This is the primary entry point for deploying Dify.

Code Reference

Source Location: docker/docker-compose.yaml, lines 697-1626

Signature:

cd docker
docker compose up -d

Import statement: Not applicable (Docker Compose CLI command).

I/O Contract

Inputs

Parameter	Type	Required	Default	Description
.env file	File	No (defaults used)	docker/.env	Environment variable overrides for all services
docker-compose.yaml	File	Yes	docker/docker-compose.yaml	Service definitions, dependencies, and network configuration
COMPOSE_PROFILES	String	No	weaviate,postgresql	Comma-separated list of profiles to activate (vector DB + metadata DB)
-d flag	CLI flag	No	N/A	Detached mode; containers run in background

Outputs

Output	Type	Description
Running containers	Docker containers	All services started according to dependency order and profile selection
Exposed ports	Network bindings	nginx on ports 80/443, plugin_daemon debug on 5003
Named volumes	Docker volumes	Persistent data for db, redis, vector store, app storage, sandbox
Docker networks	Bridge networks	default, ssrf_proxy_network (internal), milvus (if active), opensearch-net (if active)

Usage Examples

Standard deployment:

cd docker
cp .env.example .env
# Edit .env as needed (SECRET_KEY, passwords, etc.)
docker compose up -d

Check service status after launch:

docker compose ps
# NAME              STATUS          PORTS
# api               running         5001/tcp
# worker            running
# worker_beat       running
# web               running
# nginx             running         0.0.0.0:80->80/tcp, 0.0.0.0:443->443/tcp
# db_postgres       running (healthy)
# redis             running (healthy)
# sandbox           running (healthy)
# plugin_daemon     running         0.0.0.0:5003->5003/tcp
# ssrf_proxy        running
# weaviate          running

View logs for a specific service:

docker compose logs -f api
docker compose logs -f worker

Restart only the API service:

docker compose restart api

Scale Celery workers (not supported directly via replicas due to volume mounts, but possible with override):

# Adjust CELERY_WORKER_AMOUNT in .env to control concurrency within a single worker
# Or use CELERY_AUTO_SCALE=true with CELERY_MAX_WORKERS and CELERY_MIN_WORKERS

Tear down the entire stack:

docker compose down
# Add -v to also remove volumes (WARNING: destroys data)

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment