---

name: deployment-and-operations description: "Install, run, configure, deploy, and troubleshoot Langflow in local, Docker, Compose, and operational environments." disable-model-invocation: true metadata: disco-role: operating license: MIT

Deployment and Operations

Use this sub-skill when the task is to install Langflow, start or configure the server, prepare .env files, run Docker or Docker Compose deployments, preserve data while upgrading, package flows or custom code into images, or diagnose operational failures around databases, volumes, credentials, networking, logging, storage, and runtime dependencies.

Route elsewhere when the task is primarily:

• Editing Langflow source, tests, formatting, generated files, or release policy: use ../repo-maintenance/.
• Implementing REST, SDK, or application client code against a running server: use ../sdk-and-api-clients/.
• Backend route/service internals or database migration implementation: use ../backend-runtime/.
• lfx run or lfx serve stateless executor workflows: use ../executor-cli/.

Start Here

1. Read references/install-and-run.md for local package installs, source runs, CLI commands, .env loading, precedence, superuser/API-key operations, and safe upgrade checks.
2. Read references/deployment.md for Docker, Compose, PostgreSQL, volumes, custom flow images, custom Langflow images, multi-worker/Redis, reverse proxy, and observability patterns.
3. Read references/troubleshooting.md when startup, imports, optional dependencies, database URLs, migrations, credentials, Docker networking, storage, logs, or worker queues fail.
4. Validate .env files before deployment with the bundled helper:

python scripts/check_env_vars.py --env-file .env --context docker-compose

Fast Operational Checklist

• Prefer uv run langflow run from an environment where langflow is installed; use direct langflow run only when the executable is intentionally on PATH.
• Remember precedence: CLI options override .env; .env overrides system environment; Docker -e and Compose environment entries are system environment inside the container unless an --env-file or service env_file supplies values.
• For local production-like persistence, set LANGFLOW_CONFIG_DIR and back it up with the database; for Compose, mount the config directory and PostgreSQL data on persistent volumes.
• For PostgreSQL, use a full LANGFLOW_DATABASE_URL such as postgresql://user:password@postgres:5432/langflow from inside Compose, not localhost unless the database is in the same container namespace.
• For public or shared deployments, disable automatic login, set a non-default secret key, create a superuser, protect API keys, and put Langflow behind HTTPS/reverse-proxy authentication appropriate to the environment.
• For multi-worker deployments, configure all workers consistently and use LANGFLOW_JOB_QUEUE_TYPE=redis plus a shared Redis queue URL; mixed asyncio/Redis workers are not supported.

Useful Commands

uv pip install -U langflow
uv run langflow run --host 127.0.0.1 --port 7860
uv run langflow run --env-file .env --backend-only --log-level info
uv run langflow superuser --username admin --password 'change-me'
uv run langflow api-key
uv run langflow migration
docker run --rm -p 7860:7860 langflowai/langflow:latest
docker compose up -d
docker compose logs -f langflow

Run langflow migration before langflow migration --fix, and only run fix mode after backing up the database or confirming the environment is disposable.

Evidence Base

This guidance is distilled from Langflow's README quickstart, development setup guidance, CLI configuration docs, environment-variable docs, Docker deployment docs, deployment overview, Compose examples, setup scripts, and verified installed package facts for langflow, langflow-base, lfx, and langflow-sdk. Runtime instructions here are self-contained and do not require reopening those source files.