- Rename configuration.md to auth.md for clarity - Rename storage-and-blob-behavior.md to object-blob-storage.md - Split server library import into dedicated documentation page - Rename intro.md to introduction.md - Merge support.md into support-and-contributing.md - Add interactive tabs to configuration and setup guides - Improve cross-references and navigation between documentation pages - Update sidebar structure and Docusaurus config for new file paths
2.9 KiB
| title |
|---|
| Vercel Deployment |
import Tabs from '@theme/Tabs'; import TabItem from '@theme/TabItem';
This guide covers deploying OpenReader WebUI to Vercel with external Postgres and S3-compatible object storage.
What works on Vercel
- Documents (PDF/EPUB/TXT/MD) work with
POSTGRES_URL+ external S3 storage. - Audiobook routes work on Node.js serverless functions using
ffmpeg-static/ffprobe-static.
:::warning DOCX Conversion Limitation
docx conversion requires soffice (LibreOffice), which is not available in a standard Vercel runtime.
:::
1. Environment Variables
POSTGRES_URL=postgres://...
USE_EMBEDDED_WEED_MINI=false
S3_ACCESS_KEY_ID=...
S3_SECRET_ACCESS_KEY=...
S3_BUCKET=...
S3_REGION=us-east-1
S3_ENDPOINT=https://...
S3_FORCE_PATH_STYLE=true
S3_PREFIX=openreader
BASE_URL=https://your-app.vercel.app
AUTH_SECRET=...
NEXT_PUBLIC_NODE_ENV=production
NEXT_PUBLIC_ENABLE_AUDIOBOOK_EXPORT=true
NEXT_PUBLIC_ENABLE_WORD_HIGHLIGHT=true
:::tip For all variables and defaults, see Environment Variables. :::
2. FFmpeg/ffprobe packaging in Vercel functions
ffmpeg-static and ffprobe-static binaries must be included in function traces. This repo already does that in next.config.ts via outputFileTracingIncludes for:
/api/audiobook/api/audiobook/chapter/api/audiobook/status/api/whisper
:::info
serverExternalPackages should include ffmpeg-static and ffprobe-static so package paths resolve at runtime instead of being bundled into route output.
:::
If you change route paths or split handlers, update outputFileTracingIncludes accordingly.
3. Function memory sizing
FFmpeg workloads benefit from more memory/CPU. This repo includes:
{
"$schema": "https://openapi.vercel.sh/vercel.json",
"functions": {
"app/api/audiobook/route.ts": { "memory": 3009 },
"app/api/whisper/route.ts": { "memory": 3009 }
}
}
Adjust memory per route if your files are larger or your plan differs.
4. Runtime expectations and caveats
- Audiobook APIs require S3 configuration; otherwise they return
503. - For production Vercel deploys, use
POSTGRES_URLinstead of SQLite. - Filesystem-to-object-store migrations run via server scripts/entrypoint (
scripts/migrate-fs-v2.mjs), not API routes. - Vercel deployments do not run
scripts/openreader-entrypoint.mjs, so runpnpm migrate-fsin a controlled environment when migrating legacy filesystem data.
5. Smoke test after deploy
- Upload and read a PDF/EPUB document.
- Confirm sync/blob fetch works across refreshes/devices.
- Generate at least one audiobook chapter and play/download it.
- If using word highlighting, verify timestamps are produced and rendered.