Frequently asked questions

Short answers to the questions people ask before and after installing Arclight, grouped by theme. Each answer links to the page with the full steps. If your question is “something is broken”, start at Troubleshooting; for unfamiliar words see the Glossary.

About

What is Arclight?

One self-hosted app that does the jobs of Sonarr (TV), Radarr (movies), Prowlarr (indexers), Bazarr (subtitles) and Seerr/Overseerr (requests), with its own optional torrent engine and VPN integration. You get one login, one database and one web UI. See the documentation home.

Is Arclight a media server?

No. It finds, downloads, imports, renames and fetches subtitles for your TV and movies, then tells Jellyfin, Emby or Plex to rescan. You still watch through your media server. See Media servers.

Does it replace Sonarr, Radarr, Prowlarr, Bazarr and Overseerr?

Yes, each of those jobs is built in: Library (Sonarr/Radarr), Indexers (Prowlarr), Subtitles (Bazarr) and Discover and requests (Seerr). It also manages music (Lidarr) and audiobooks (Readarr) in the same Library. It does not replace a media server (the known gaps are in known limitations).

Can I run it next to my existing Sonarr and Radarr while I try it?

Yes, but let only one app download, or they will fight over the same titles. Radarr also uses port 7878, so give Arclight another host port (ARCLIGHT_PORT, or the left side of the 7878:7878 mapping in compose). See Migrating.

Can I import my existing setup?

Yes. System > Import (“Import your existing stack”) reads Sonarr, Radarr, Prowlarr, Bazarr, Overseerr / Jellyseerr and qBittorrent, through their live API or their files, and lets you choose what to copy. See Migrating.

Will importing from qBittorrent make my torrents download again?

No. Arclight checks that the files are at the mapped path first, refuses any torrent whose files are missing or the wrong size, and either hashes the existing data or trusts complete files, so a complete torrent comes up seeding. Upload totals, seeding time and limits carry over, so private-tracker rules keep counting. Files are never moved or deleted. See Adopting qBittorrent torrents.

Does the import change or delete anything in Sonarr, Radarr and the others?

No. The old apps are only read, and nothing in Arclight is ever deleted by the import. The one optional exception is the clean-up on the report, which can pause torrents in qBittorrent or remove them from its list (files kept) after you confirm it.

Can I undo an import?

Yes. A real run takes an Arclight backup first; restore it from System > Backups. Restoring does not touch qBittorrent. See Dry run, backup and rollback.

Should I use the live API or the files?

Use the API while the old app runs, and files when it is stopped or gone. Recent Sonarr, Radarr, Prowlarr and Bazarr hide passwords and API keys over the API, so those need re-entering; files carry them. See API or files.

Can I run the import twice?

Yes. Everything already imported is skipped, so repeating it never duplicates anything. A run that failed or was interrupted can be continued with Resume.

Will my Overseerr users keep their passwords?

No password is copied. Imported users sign in with their Jellyfin, Emby or Plex account, and an admin can set a local password. The import never creates an Arclight administrator. See Overseerr and Jellyseerr.

What does it cost, and what is the licence?

It is free software under the GPL-3.0 licence (LICENSE). It is an independent implementation; NOTICE.md lists the projects whose behaviour informed it.

Install

What do I need to install it?

Docker is the supported way to run it, plus a free TMDB API key and one data folder that holds your downloads and your TV and movie libraries. See Requirements and Install with Docker.

Can it run without Docker?

Technically yes: it is a Node 24 app, and the repository includes a native Windows route (scripts/local-test.ps1) that the user guide describes as meant for trying Arclight, development and testing. For a long-running server use Docker. See Install on other platforms.

Does it work on my NAS or computer?

Any machine that can run Docker Compose can run it. Pick the guide for your platform; each one ends with the same folder layout and first-run steps.

Platform	Guide
Generic Linux server	Install with Docker
Synology	Install on Synology
UGREEN NAS	Install on a UGREEN NAS
Unraid	Install on Unraid
TrueNAS SCALE	Install on TrueNAS SCALE
QNAP	Install on QNAP
Windows	Install on Windows
macOS	Install on macOS
Raspberry Pi	Install on a Raspberry Pi
Portainer (any host)	Install with Portainer

Set PUID and PGID to the user that owns your data folder.

Which install guide should I follow?

Use the one that matches the machine that will run it. If your NAS is not listed but runs Docker Compose, use the generic Linux guide. If you already manage containers through Portainer, use the Portainer guide. Running it somewhere you can reach from outside your home is a separate step: see Can I expose it to the internet?.

Which ports does it use?

The web UI and API use 7878 (change it with ARCLIGHT_PORT). The built-in engine’s peer port defaults to 6881 (TCP, plus UDP for DHT), though a port forwarded by the VPN takes precedence. If you opt in to qBittorrent, its web UI is on 8080. FlareSolverr (8191) and the Gluetun control server (8000) are only used inside the stack, and Whisper (9000) only if you enable it. See Install with Docker.

Do I need a TMDB key, and can I use a free one?

You need one, and the free key from themoviedb.org (Settings > API, the v3 key) is what Arclight expects. Titles, posters and episode data all come from TMDB. You enter it in the first-run wizard or later in Settings > General, or you can set ARCLIGHT_TMDB_API_KEY.

Which languages is the interface available in?

English only for now. German, French, Spanish and Dutch files exist, but their strings are untranslated placeholders (each starts with a marker such as [de]), so choosing them shows English text with markers. Help is welcome: see Translating Arclight. The language picker is under Settings > General.

Can I use it on my phone?

Yes. The layout collapses to a menu drawer below 900 px wide, and Arclight ships a web app manifest and a service worker, so a browser can install it to your home screen. Browser push notifications need HTTPS (or localhost); on iPhone and iPad add it to the home screen first. See Notifications.

VPN and privacy

Does it need a VPN?

No. The VPN is off by default, and there is a no-VPN compose layout. If you download torrents you almost certainly want one; usenet does not need one. See VPN.

Which VPNs are supported?

The shipped compose files use Gluetun with ProtonVPN over WireGuard, and Arclight reads Gluetun’s status and forwarded port. Natively on Windows there are two other providers: WireGuard for Windows (needs an elevated process) and interface binding for a VPN you run yourself. Other VPN clients are not integrated. See VPN.

Can I use a VPN provider other than Proton?

Yes. The shipped compose files are set up for Proton, but Gluetun supports many providers; edit the Gluetun service environment to use yours (see the Gluetun documentation for your provider’s settings). Port forwarding depends on the provider: check that yours offers it, because without a forwarded port you will be able to download but may get fewer incoming connections. Arclight only reads Gluetun’s status and forwarded port, so the rest of the setup is unchanged. See VPN.

What does the kill switch do?

With the Gluetun layout, Arclight and the torrent engine share Gluetun’s network, so Gluetun’s firewall blocks everything that is not going through the tunnel. On top of that, “Pause grabs when down” (on by default) stops new grabs while the VPN is down or leaking, and the engine’s “Bind to VPN” (on by default) pauses torrents until it is back. See VPN and paused by the VPN kill switch.

Does Arclight phone home or collect telemetry?

No analytics or telemetry are in the code. It makes outbound requests only for things it needs, listed in the next answer, and your own indexers, clients and notification targets. See Users and security.

Exactly what does it contact on the internet?

TMDB (metadata, and your browser loads posters from TMDB’s image server); GitHub and the GitHub container registry for the update check (anonymous, can be switched off in Settings > Updates); the project’s GitHub repository for the indexer definition feed (on by default, every 12 hours, can be set to manual or off in Settings > Indexer Options); and, only if you enable them, the TRaSH guides repository for guides sync, a public IP check service such as api.ipify.org for native VPN providers, subtitle providers, import lists and notification services. See Updating.

Can I stop it checking for updates?

Yes. Turn off update checks in Settings > Updates; nothing else about Arclight depends on it. See Updating.

Downloading

Do I need qBittorrent?

No. Arclight downloads torrents with its own built-in engine, which is already set up in every supplied Docker file (a client called Built-in is added on first start), so you can go straight from the setup wizard to downloading. The built-in engine has some gaps compared with qBittorrent (see known limitations); keep or add qBittorrent if you depend on those. Transmission, Deluge, rTorrent, SABnzbd and NZBGet are also supported, and you can have several clients at once. See Download clients.

Can I keep qBittorrent?

Yes, it is still supported. In the Docker files it sits behind the compose profile qbittorrent: run docker compose --profile qbittorrent up -d to start it, then add it under Settings > Download Clients with host localhost (VPN layouts) or qbittorrent (no VPN), port 8080, and set its default save path to /data/downloads. If you already run your own qBittorrent, add that instead: give Arclight its own category (such as tv) so it can tell its downloads from yours, and press Test. If it runs behind Gluetun, Arclight can also push the VPN’s forwarded port into it. If you run it next to the built-in engine without a VPN, give them different peer ports. See Download clients.

Does it support usenet?

Yes. Add Newznab indexers under Settings > Indexers and a SABnzbd or NZBGet download client. The usenet clients have only been tested with stubs, not end to end against a real server, so please report issues. See Indexers and Download clients.

Which indexers can I use?

Torznab and Newznab endpoints (for example from Jackett or a usenet provider), plus built-in site definitions for a few public trackers. Those definitions have not been verified against the live sites, and public sites move often. Cloudflare-protected sites need FlareSolverr. See Indexers and Cloudflare challenges.

Why isn’t an episode or movie being searched?

Probably because it is not released yet. Automatic searches and the Search buttons skip unreleased items; they show under Wanted > Upcoming. Interactive search still works, and Search unreleased items in Settings > Media management lets your own searches include them. See Wanted.

Why is nothing downloading?

Check Settings > Download Clients > Test, whether the VPN is paused, and that the release has seeders. See download stuck, no peers and nothing found.

Can the torrent engine run on a different machine?

Yes, a seedbox or VPS can run the engine and the main Arclight drives it through a Remote Engine client. See Remote engine.

Library and quality

Why hardlinks, and what do I need for them?

A hardlink makes the imported file appear in your library without a second copy, so the torrent keeps seeding and uses no extra space. It only works when the downloads folder and the library are on the same filesystem and, in Docker, under one mount. The compose files mount a single share, ${DATA_ROOT}:/data, with /data/downloads, /data/media/tv and /data/media/movies inside it. The switch is Settings > Media Management > “Use hardlinks instead of copy” (on by default). See Folders and hardlinks and hardlinks not working.

Why does the compose file mount one /data folder instead of separate /tv, /movies and /downloads?

Linux cannot create a hardlink between two different mounts, even when both folders sit on the same physical disk. With separate /tv, /movies and /downloads mounts, every import falls back to a full copy: slower, and it uses double the space while the torrent keeps seeding. With one /data mount, imports are instant hardlinks and seeding keeps working. Arclight’s root folders are /data/media/tv and /data/media/movies. See Folders and hardlinks.

Imports are slow or use double the disk space. Why?

Arclight is copying instead of hardlinking, almost always because downloads and media are not under one mount, or because the folders are on different filesystems. Check that your compose file has the single ${DATA_ROOT}:/data mount and that your download client saves under /data/downloads. See hardlinks not working and hard links not possible.

I installed with three mounts (/tv, /movies, /downloads). How do I migrate?

Move your folders under one parent, switch the compose file to the single ${DATA_ROOT}:/data mount (and remove TV_ROOT, MOVIES_ROOT and DOWNLOADS_ROOT from .env), then change your Arclight root folders to /data/media/tv and /data/media/movies and re-point your series and movies. The full steps are in Folders and hardlinks.

If I use qBittorrent, what save path should it use, and do I need a remote path mapping?

In qBittorrent (optional) go to Options > Downloads and set the default save path to /data/downloads. Because Arclight and qBittorrent both see the same /data mount, the paths match and no remote path mapping is needed. The built-in torrent engine defaults to /data/downloads when /data exists; older installs that already use /downloads keep it. See Download clients and Folders and hardlinks.

Can I keep 4K and 1080p of the same title?

Yes, two ways. On a series or movie page, add Versions (each with its own tag and quality profile; the editor allows up to 8) and Arclight searches and upgrades each version separately. For requests, turn on “Enable 4K requests” in Settings > Requests; 4K requests use their own root folder and profile so both copies live side by side. See Library and Discover and requests.

What are quality profiles and custom formats?

A quality profile says which qualities you accept and when to stop upgrading; custom formats score releases on things like codec or release group. See Quality profiles.

Can I sync TRaSH guides?

Yes, guides sync is off by default. Enable it under Settings > Guides, map guide profiles to your own, and choose whether it only notifies you or applies changes. See Quality profiles.

Does it handle subtitles?

Yes, it replaces Bazarr, with providers such as OpenSubtitles.com and SubDL, optional Whisper (you provide the service), and a language profile per title. Some providers are best-effort or untested against the live service. See Subtitles.

Does Arclight support hardware transcoding?

Yes, for transcode rules (which are optional). They can encode on an Intel iGPU or Arc (Quick Sync, VA-API), an AMD GPU (VA-API) or NVIDIA (NVENC). Pass /dev/dri or the NVIDIA runtime into the container with the commented block in the compose file; the image already bundles the Intel drivers (amd64 only). NAS models with no integrated GPU encode on the CPU. Settings > Transcode > Hardware acceleration tests each backend and shows why one is unavailable. See Hardware transcoding.

Do other apps work with it, like LunaSea or Recyclarr?

Arclight serves a Sonarr and Radarr v3 compatible API under /api/v3, and API.md lists Overseerr/Jellyseerr, Bazarr, Prowlarr, Notifiarr, Homarr, Recyclarr, LunaSea, nzb360 and dashboards as tools that can point at it. Recyclarr can sync quality profiles but not every setting, and the calendar is episode-only (so movies may not show in Radarr-only apps). Tools not on that list are untested. Details are in API.md and known limitations.

Requests and users

Can my family use it without admin rights?

Yes, and that is the default for everyone but you. Accounts with the role user get the request portal only: Discover, their own requests and issues, and their account. They never see the dashboard, the library, Activity, Settings or System, and the api refuses those to them (reads included). If you want someone to see more, give them read-only app access (Library, Calendar, Activity, Wanted) under Settings > Security > Users > Edit. Fine-grained request rights (request, 4K, auto-approve) are permission switches per user. See Discover and requests and Users and security.

Yes, when a media server is configured with sign-in enabled, the login page offers it, and imported media-server accounts become Arclight users who sign in with their media-server password. Those accounts are always users with the request portal. See Media servers.

Can a Jellyfin user become admin?

No, not through Jellyfin, Emby or Plex. Signing in with a media server only ever creates a user, even for an administrator of that media server, and an admin account can never sign in through a media server or have one linked to it. Admins sign in with their Arclight username and password. An admin can promote a user, but only after the account has an Arclight password (the editor asks for one), and the promotion signs that person out everywhere.

Do requests need approval?

That is up to you. Users with an auto-approve permission are approved automatically; everyone else’s requests wait for an admin in Requests. Quotas and approval rules are in Settings > Requests. See Discover and requests.

Can people ask for subtitles or another audio language?

Yes. With the Request subtitles & languages permission (included in Request), a Request subtitles or audio button appears on titles in the library. Subtitle requests are approved automatically by default; audio requests wait for a manager. See Discover and requests.

Updates and data

How do updates work?

Arclight checks GitHub for new releases (every 24 hours by default, stable channel) and shows a notice on Settings > Updates and System > Updates. By default it only notifies you; update by pulling the new image, or mount the Docker socket and use “Update now” or the automatic mode (3:00 to 6:00 by default) for containers labelled arclight.autoupdate=true. It migrates its own database when it starts, so take a backup first. See Updating.

Where is my data?

In the config folder: /config in Docker (the arclight folder under CONFIG_ROOT in the compose files), ./config natively, or whatever ARCLIGHT_CONFIG points to. It holds the SQLite database arclight.db, logs/, backups/, and torrents/ if you use the built-in engine. Your media stays in your own folders. See Backup and restore.

How do I back up?

Backups run daily at 04:00 by default (keeping 7, and deleting automatic ones older than 28 days) and are listed under System > Backups, where you can also create, download, upload and restore. See Backup and restore.

How do I move to a new machine?

Back up (System > Backups), install Arclight on the new machine, then restore the backup there and restart. Keep the same media paths, or fix them afterwards. Copying the whole config folder works too. See Migrating.

Security

How do I reset the admin password?

Another admin can do it under Settings > Users > Reset password, and any user can change their own on their profile page. If you lost the only admin password there is no command-line reset: restore a backup from before the change, or stop Arclight, move arclight.db aside and run the first-run wizard again (you lose settings, not media). Advanced: your API key grants admin rights, so PUT /api/v1/user/<id> with a new password also works. See locked out.

Can I expose it to the internet?

Only behind HTTPS with sign-in on; never turn the method to None for an internet-facing install. Reverse proxy, Cloudflare Tunnel and Tailscale are walked through in Exposing Arclight outside your home. Failed logins are rate limited (10 per address per 15 minutes), and there is no two-factor sign-in. See Users and security and Cloudflare.

What is the “None on the local network” option?

In Settings > Security, the authentication Method can be a login page, a browser prompt (HTTP Basic), or None on the local network. With the last one, only loopback and private-range addresses skip sign-in; public addresses always need credentials. Behind a reverse proxy that does not forward X-Forwarded-For, every visitor can look local. See Users and security.

In Settings > Security. There is one key per install and it grants admin access, so give it only to apps you trust (Prowlarr, Recyclarr, dashboards). Regenerating it breaks every app using the old one. See Users and security.

Performance on a NAS

How much RAM and CPU does it need?

The image is about 80 MB to pull and about 220 MB on disk, and answers HTTP within about two seconds of starting. The project’s own smoke test expects the process to stay under roughly 1 GB of RAM when idle with 500 titles; that figure comes from the checklist, not a guarantee. See Requirements.

Will it slow my NAS down?

Background work is scheduled (RSS sync, refreshes, backups; see System > Tasks) and the built-in engine’s limits are adjustable under Settings > Torrent Engine (connections, download and upload limits). Optional extras such as Whisper transcription and transcode rules are the heavy parts, and both are opt-in. See System and Download clients.

Why is the first search slow?

Cloudflare-protected indexers go through FlareSolverr, which opens a browser for each solve, and indexers are rate limited individually. See Cloudflare challenges and slow engine start.