Skip to content

no-docker-quickstart.md

Latest commit

 

History

History
78 lines (49 loc) · 2.65 KB

no-docker-quickstart.md

File metadata and controls

78 lines (49 loc) · 2.65 KB
Raw

No-Docker Setup

Docs → MemSense Docs See also: Local BGE One-Click Deployment · API Smoke Test

What this page is for

This page covers the advanced setup path for macOS / Linux environments where Docker is unavailable. The main README uses Docker as the recommended path because it brings up Postgres, the API server, workers, and local BGE together.

Windows no-Docker setup is still being tested. Windows users should use Docker Desktop and the PowerShell commands in the main README.

Prerequisites

  • Node.js 20+ and npm
  • PostgreSQL 16+ with pgvector
  • Python 3.11+ with venv support for local BGE embedding

On macOS, the bootstrap script can install PostgreSQL and pgvector through Homebrew. On Linux, install PostgreSQL, pgvector, Python, and venv support with your system package manager first.

Choose an embedding mode

Run one setup path. bootstrap-nodocker.sh is the first-time setup step; start-bash.sh starts the runtime processes after setup.

Local embedding

bash scripts/bootstrap-nodocker.sh local
bash scripts/start-bash.sh

This installs dependencies, initializes the database, prepares the local BGE Python service, and starts the server, embedding worker, tag worker, and BGE service.

OpenAI-compatible embedding

Set MEMSENSE_OPENAI_API_KEY in .env, then run:

bash scripts/bootstrap-nodocker.sh openai
bash scripts/start-bash.sh

This installs dependencies, initializes the database, and starts the server, embedding worker, and tag worker.

Runtime controls

# Start after the first bootstrap, after reboot, or after a manual stop
bash scripts/start-bash.sh

# Stop the local background processes when you are done or before changing .env
bash scripts/stop-bash.sh

# Restart already-running bash-managed processes in place
bash scripts/start-bash.sh --restart

start-bash.sh starts the local API server, embedding worker, tag worker, and the BGE service in local embedding mode. It expects .env and Node dependencies to exist, so run bootstrap-nodocker.sh first on a fresh checkout.

stop-bash.sh stops the processes started by start-bash.sh and clears .runtime pid/log files. Use it before changing ports or providers in .env, before switching back to Docker, or when you no longer need the local service.

Logs:

  • .runtime/server.log
  • .runtime/worker.log
  • .runtime/tag-worker.log
  • .runtime/bge.log (if local BGE mode)

Next pages