Second step of ADR-008. A session spawns ffmpeg with the detected encoder's args, packages the file as HLS, and is reused while it is alive so two range requests do not start two ffmpegs. Three things it has to get right, and all three are about not burning the GPU: a session nobody is watching is reaped after a few minutes idle; concurrent sessions are capped rather than queued, because a GPU encodes one or two streams and a queue would just make everyone wait longer; and Shutdown stops everything so a restart does not leave ffmpeg running. Found a real bug in review, and it is the kind that only a real host catches. The hardware args were placed BEFORE -i. ffmpeg reads options before the input as INPUT options, so "-c:v h264_nvenc" was looked up as a decoder and failed with "Unknown decoder 'h264_nvenc'". Transcoding would have produced nothing on any NVENC host while looking correct in the code. Verified against the real hardware rather than reasoned about: the same ffmpeg command produces zero segments with the args before -i and a valid playlist and segments with them after. The test now asserts the ordering directly and names the reason, so a future reordering fails with an explanation instead of two arg lists. Red-green: moving them back fails with "the hardware args must follow the input". Verified end-to-end on this host: the Manager started a session against a real 20-second file and produced a 132-byte EVENT playlist and segments. Known limit: a stopped session leaves its segment directory on disk, so the cache directory needs pruning out of band. |
||
|---|---|---|
| cmd/vellum | ||
| docs | ||
| internal | ||
| migrations | ||
| scripts | ||
| web | ||
| .dockerignore | ||
| .gitignore | ||
| compose.yml | ||
| DEPLOY.md | ||
| Dockerfile | ||
| entrypoint.sh | ||
| go.mod | ||
| go.sum | ||
| LICENSE | ||
| Makefile | ||
| README.md | ||
Vellum
One application, one process, one database for every media type — ebooks, comics, audiobooks, music, photos, games, and video — behind a single account system with real per-user progress and per-library permissions.
A modular monolith. One Go binary, one SQLite file, one web frontend.
The full plan lives in ~/.opencode/plan/vellum.md.
Status
| Phase | Scope | State |
|---|---|---|
| 0 | Repo, config, migrations, health, Docker | done |
| 1 | Users, sessions, library grants | done |
| 2 | Scanner, items, assets | done |
| 2b | FTS5 search | done |
| 3 | EPUB reader + progress | done |
| 4 | CBZ comic reader | done |
| PDF reading (pdf.js) | done | |
| 5 | Job queue, metadata (Open Library, Google Books, MusicBrainz, ComicVine, IGDB, TMDB) | done |
| 6a | Audiobooks + music playback | done |
| 6c | Photos + thumbnails | done |
| 7 | Games: ROM scanning, platforms, cloud saves, EmulatorJS | done |
| UI | Svelte app: home, browse, item pages, readers | done |
| 6b | Music smart playlists | not started |
| — | Video (stays on Jellyfin) | deferred |
What works today: an administrator adds libraries pointing at server folders, scans them, and optionally enriches metadata. Users sign in and read EPUBs and PDFs, read CBZ comics, listen to audiobooks track-by-track, browse photos, and play emulated games in the browser with their own cloud saves — with position saved per account, per file.
Six media types work: ebooks, comics, audiobooks, music, photos, games. Video is the only one not implemented, and it stays on Jellyfin on purpose.
Of the services in the existing stack, this covers Stump, Kavita, Audiobookshelf, Navidrome, and RomM. Only Jellyfin (video) is left.
Quick start
sudo dnf install golang # once; Go is not in the base image here
make build
make run
curl localhost:7788/healthz # {"database":"ok","status":"ok",...}
Run with media roots configured:
VELLUM_MEDIA_DIRS=/mnt/storage/media/ebooks:/mnt/storage/media/comics make run
Configuration
All configuration is environment variables. Every one has a default, so the binary runs with no configuration at all.
| Variable | Default | Meaning |
|---|---|---|
VELLUM_LISTEN |
:7788 |
Bind address |
VELLUM_DATA_DIR |
./data |
SQLite database lives here |
VELLUM_CONFIG_DIR |
./config |
Regenerable cache (covers, pages) |
VELLUM_MEDIA_DIRS |
(none) | Read-only library roots, colon-separated |
VELLUM_ADMIN_USER |
(none) | Creates the first admin, only on an empty database |
VELLUM_ADMIN_PASSWORD |
(none) | Password for that first admin (min 8 chars) |
VELLUM_COOKIE_SECURE |
false |
Set true when serving over TLS |
VELLUM_COMICVINE_API_KEY |
(none) | Enables comic metadata via ComicVine |
VELLUM_DISCOVERY |
on |
Set off to stop advertising over mDNS on the LAN |
VELLUM_IGDB_CLIENT_ID |
(none) | Enables game metadata via IGDB |
VELLUM_IGDB_CLIENT_SECRET |
(none) | IGDB client secret (both required) |
VELLUM_TMDB_API_KEY |
(none) | Enables movie and TV metadata via TMDB |
VELLUM_LOG_LEVEL |
info |
debug, info, warn, error |
VELLUM_LOG_FORMAT |
text |
text or json |
VELLUM_SESSION_TTL |
720h |
Login session lifetime |
VELLUM_SHUTDOWN_GRACE |
15s |
Graceful shutdown timeout |
The first admin is created only when no users exist, so setting these on an established install does nothing. If the database is empty and they are unset, the server logs a warning and refuses to invent a default password.
Layout
cmd/vellum/ main — wiring and lifecycle only, the one place packages meet
internal/
core/ config, logging, database, migrations, timestamps
httpx/ shared JSON helpers
httpapi/ router, middleware, security headers, static SPA serving
auth/ argon2id, sessions, grants, bootstrap
apitoken/ credentials for non-browser clients (Jellyfin, Subsonic)
catalog/ THE data seam: read an item, and check you may
library/ library CRUD and the media-type registry
scanner/ walk the filesystem and reconcile with the database
probe/ ffprobe wrapper — codec facts (optional dependency)
media/ generic item/file/cover/progress endpoints
home/ the five home shelves
search/ FTS5 queries
metadata/ the enrichment job and its matching policy
providers/ Open Library, Google Books, MusicBrainz, ComicVine, IGDB, TMDB
playback/ playback engine selection endpoint
engine/ playback engine registry (html5, mpv, vlc, mdk)
progress/ the polymorphic per-user progress store
stream/ range-capable file serving
jobs/ table-backed job queue
natsort/ natural ordering for pages and tracks
jellyfin/ Jellyfin protocol adapter (+ dto/)
subsonic/ Subsonic protocol adapter (+ dto/)
discovery/ mDNS advertisement on the LAN
arch/ import-rule tests; holds no code
modules/
ebooks/ EPUB parsing and cover extraction
comics/ CBZ page listing and extraction
games/ platforms, IGDB, cloud saves, EmulatorJS
photos/ thumbnail generation
web/ Svelte app; dist/ is embedded via web/embed.go
migrations/ embedded SQL, applied in order at boot
docs/ architecture guide and decision records
Module rule: the core never imports a feature module; modules import the core and mount their own routes, tables, and scanner rules. Adding a media type is a new package plus a migration, never a core edit.
Layer rule: packages are arranged in six layers and a package may never
import from a higher one. This is enforced by tests in internal/arch, not by
convention — see docs/ARCHITECTURE.md.
Schema rule: media items are global; everything user-specific is a join. Nothing may hold per-user state on a media row.
Documentation
- docs/ARCHITECTURE.md — how the pieces fit and why.
- docs/decisions/ — the decisions that were expensive to reverse, with the alternatives that were rejected.
Database
SQLite in WAL mode, opened through modernc.org/sqlite — a pure-Go driver, so
the binary builds with CGO_ENABLED=0 and cross-compiles without a toolchain
per target.
Migrations are embedded and applied in filename order; applied files are
recorded only by number in PRAGMA user_version, so never edit an applied
migration — add a new one.
Back up by copying the database (or its directory); VACUUM INTO 'backup.db'
takes a consistent snapshot while running.
Deploy
make docker # podman build
podman compose up -d # see compose.yml
The container mounts /data (database), /config (cache), and /media
(read-only). On SELinux hosts every bind mount needs :z — compose.yml
already has it.
The application never writes to /media.
Development
make dev # backend + Vite dev server together, with hot reload
make build # builds the frontend, then the Go binary with it embedded
make build-go # Go only, when the frontend is unchanged
make check # gofmt + vet + test
make run # run the server on :7788
cd web && pnpm dev # frontend only; needs the backend running on :7788
make dev is the one to reach for while developing: it builds and starts the
backend, waits for it to answer, then starts Vite. Vite proxies /api to the
backend so the browser sees a single origin and sessions behave exactly as in
production. Ctrl-C stops both.
If you run Vite on its own, the backend must already be running, or every
request fails with ECONNREFUSED on :7788.