Fresh install fails: DB tables never created in production mode

[Thoroslives]

Bug

Fresh installs with the Docker image (ghcr.io/navino16/librarr:latest or :develop) fail with SqliteError: no such table: user on every route. The setup wizard returns Internal Server Error when attempting to create an admin account.

Root Cause

In server/datasource.ts, the synchronize option is:

synchronize: isProduction ? false : synchronize,

When NODE_ENV=production (set by the Dockerfile), synchronize is always false regardless of DB_SYNCHRONIZE. The code falls through to use migrations instead, but no migration files ship with the image:

$ find /app/dist/server/migration -type f
(empty)

This means a fresh install in production mode can never create its database tables.

Steps to Reproduce

1. docker compose up -d with the docker-compose from the docs
2. Open http://localhost:5055
3. Complete step 1 (admin account) → Internal Server Error
4. Container logs show: SqliteError: no such table: user

Workaround

Temporarily set NODE_ENV=development and DB_SYNCHRONIZE=true to let TypeORM create the tables, then switch back to NODE_ENV=production (keeping DB_SYNCHRONIZE=true). The app crashes under NODE_ENV=development because Next.js standalone mode requires production, so you must switch back after the DB initializes.

Suggested Fix

Either:

1. Ship migration files with the image so typeorm can run them on first start
2. Change the synchronize logic to allow initial schema creation in production (e.g., sync if the DB is empty)
3. Remove the isProduction guard: synchronize: synchronize — let DB_SYNCHRONIZE control it directly

Environment

• Image: ghcr.io/navino16/librarr:latest and :develop
• Docker, Unraid host
• Node 24 Alpine base

[Navino16]

Sorry for missing this! Fixed in #14. Fresh Docker installs now auto-create DB tables via an initial TypeORM migration (migrationsRun: true in production). Existing installs are handled gracefully — the migration detects existing tables and skips creation.