If you've spent any time in homelab or self-hosting circles, you've seen people mention their "Arr stack" — often written *arr, where the asterisk is a wildcard for Son*arr, Radarr, Lidarr*, and friends. It sounds like arcane jargon, but the idea is simple: a set of apps that turn "I want to watch this" into an organized file in your library without you touching a download client.
This post explains what each piece does and how they fit together — the concepts, no setup yet.
A note on use: the arr stack is just a tool. Point it only at content you have the legal right to download where you live — your own media, public-domain works, and Linux ISOs are the intended use.
The problem the Arr stack solves
Managing a media library by hand means searching a dozen indexers, squinting at release names to guess which is good, downloading, renaming the file into something your media server understands, moving it into the right folder, and refreshing the library. Then doing it again for the next episode. It starts as a hobby and quietly becomes a part-time job.
The Arr stack automates every step. You tell it what you want, and it handles the searching, grabbing, renaming, and filing, then tells your media server to rescan so the title just appears, ready to watch.
One thing I learned running this myself: "hands-off" holds right up until the auto-downloading quietly stops. Nothing errors — the library just stops growing, and because everything still looks fine, you don't catch it for a while. Build it so you'll notice when one link stops handing off to the next.
The players
Each app is a focused piece of the pipeline. All are open-source and part of the Servarr family, so they share a near-identical interface.
| App | Handles | Thinks in terms of |
|---|---|---|
| Sonarr | TV shows | series → seasons → episodes |
| Radarr | Movies | single films + editions |
| Prowlarr | Indexers | the sources everything searches |
| A download client (e.g. qBittorrent) | The actual transfer | torrents / Usenet |
| A media server (Plex / Jellyfin) | Playback | libraries you stream |
Sonarr and Radarr are essentially the same program with different logic — Sonarr understands episode numbering and air schedules, Radarr understands releases and cuts. There are siblings too: Lidarr (music), Readarr (books, now retired). Most media setups only need Sonarr and Radarr.
How Prowlarr changed the game
An indexer is a searchable catalog of available releases — a torrent tracker or a Usenet index. Before Prowlarr, you added every indexer to Sonarr and Radarr separately, kept two credential sets current, and repeated it every time an indexer changed.
Prowlarr is the indexer manager: you add each indexer once, and it syncs it into every connected Arr app automatically. Prowlarr stores the indexer definitions; Sonarr and Radarr register as "applications" and receive the synced indexers.
Prowlarr decides where to search; Sonarr and Radarr decide what to grab. Prowlarr never downloads anything — it just gives the others a shared, always-current list of places to look.
The pipeline, end to end
You add a title (or an import list adds it automatically)
|
v
Radarr / Sonarr -- decides it's "wanted", asks Prowlarr's indexers
|
v
Prowlarr indexers -- return candidate releases, scored by quality rules
|
v
Download client -- grabs the chosen release (torrent behind a VPN)
|
v
Radarr / Sonarr -- imports: renames + moves the file into the library
|
v
Media server -- rescans; the title appears, ready to stream
Every arrow is automatic. The only human input is the first line — and even that can be automated with import lists that pull from a watchlist or "popular" feeds.
Where quality rules fit in
The Arr apps don't grab the first result they find. Each uses a quality profile — your rules for resolution, source, file size, and codecs — to score candidates and pick the best fit. The community-maintained TRaSH Guides are the reference for tuning these.
What about a VPN?
Only one piece needs one. Sonarr, Radarr, and Prowlarr move metadata and searches — no swarm traffic. The download client is what joins torrent swarms and exposes your IP, so the hardened setup routes only the torrent client through a VPN container with a kill-switch, leaving the rest on your LAN.
The wider *arr family
- Bazarr watches your Sonarr and Radarr libraries and fetches subtitles automatically.
- Lidarr does for music what Sonarr does for TV; Whisparr handles adult content.
- For requests, Jellyseerr gives the household a streaming-style page that hands approved titles to Radarr/Sonarr.
- One to skip: Readarr (books) was retired in 2025 when its metadata source broke.
Where to go next
I keep the full, continually-updated versions of these guides — deploy with Docker Compose, automation, subtitles, music, and requests — on my homelab site, documented from a working 4-node Proxmox cluster:
- What Is the Arr Stack? (full guide)
- Deploy the Arr Stack with Docker Compose
- Automating the Arr Stack
- Bazarr: Automatic Subtitles · Lidarr: Music · Jellyseerr: Requests
Originally published on peira.dev.
Top comments (0)