A seedbox that you drive by hand is a remote download folder. A seedbox with a working automation chain is infrastructure: things arrive, get named correctly, land where they belong, and appear in your media library without you touching anything.
The chain has four links. Understanding what each one does — and where they break — is most of the work.
The four pieces
The flow runs one way: a manager asks the indexer proxy what exists, picks a result, hands it to the client, waits for completion, then renames and moves the file into the library the media server watches.
Every seedbox host in our provider comparison ships this stack preinstalled. You are configuring it, not building it — which is exactly the point of a managed appliance rather than a VPS you administer yourself.
Wiring it up, in the right order
Order matters. Each step depends on the one before it, and configuring out of order is where most of the frustration comes from.
1. Client first. Confirm the torrent client works on its own and note two paths: where in-progress transfers live, and where completed ones land. Everything downstream depends on those two directories.
2. Indexer proxy second. Add your sources to Prowlarr (or Jackett), and test each one from inside the tool. A source that fails here will fail silently later, and you will spend an evening blaming Sonarr.
3. Managers third. In Sonarr or Radarr, connect the indexer proxy, then the download client. Set the root folder to your final library path — not the download path. This is the single most common misconfiguration: pointing the library at the download directory means files never get renamed and the media server sees a mess.
4. Media server last. Point Plex, Emby or Jellyfin at the library folders and enable scanning on change. Only after the rest works.
5. Then test one item end to end before adding a hundred. Watch it search, download, rename and appear. A chain that works once works a thousand times; a chain you never tested breaks in a way that is tedious to unpick.
Hardlinks: the setting that saves your disk
This is the detail that separates a setup that runs for a year from one that fills up in a month.
When a manager "moves" a completed download into your library, it can either copy the file or create a hardlink. A copy doubles the space used — you now hold the file twice, once seeding and once in the library. A hardlink is a second name for the same data on disk, costing nothing.
Hardlinks only work within the same filesystem. That means your download directory and your library directory must live on the same volume. Get the paths right and a 40 GB file occupies 40 GB while both seeding and playing. Get them wrong and it occupies 80 GB, and you will be buying storage you did not need.
Check this before you fill the disk, not after. On a 1 TB plan the difference is the whole plan.
Archiving off the box with rclone
Seedbox storage is a working tier, not an archive. It is a single host with no durability guarantee — the distinction we cover in seedbox versus cloud storage.
rclone ships in most seedbox stacks and moves data to object storage on a schedule:
# Push the library to a bucket
rclone sync /home/user/media remote:archive/media \
--transfers 4 --checkers 8 --fast-list --log-level INFO
# Encrypt before it leaves the box if the content is sensitive
rclone sync /home/user/media crypt:archive/media
Run it from cron nightly. Use sync only once you are certain of the direction — it makes the destination match the source, including deletions. copy is the safer default while you are still testing.
Where these setups break
Permissions. The download client and the manager run as different users often enough that the manager cannot move completed files. The symptom is downloads finishing and nothing appearing in the library. Check ownership on both directories first.
Category mismatch. Managers use download-client categories to identify their own transfers. If the category in Sonarr does not match the one the client applies, Sonarr never notices completion and the download sits there indefinitely.
Path mapping. The client's view of a path and the manager's view must agree. On a shared seedbox they usually do; when they do not, this is the cause of nearly every "download completed but nothing happened" report.
Seeding ratios. Managers can remove completed downloads. If your tracker requires seeding, that removal costs you ratio. Configure retention deliberately rather than discovering the policy afterwards.
Disk full mid-transfer. Set a minimum free space threshold in the client. A full disk corrupts in-progress transfers and takes the whole chain down.
Frequently asked questions
Do I need Prowlarr if I already have Jackett?
No. Prowlarr is the more actively developed successor and pushes configuration to the managers automatically rather than requiring manual entry, but Jackett works. Do not run both.
Why is my download stuck at completed?
Almost always category mismatch, permissions, or path mapping. Check them in that order.
Can I automate on a shared seedbox slot?
Yes — this is precisely what the preinstalled stack is for. You do not need root, and you do not need a VPS.
How much storage do I need?
Less than you think if hardlinks are configured, roughly double if they are not. Get that right before sizing the plan.
Is this legal?
The tooling is entirely legal and widely used for software distribution and dataset transfer. What you move is governed by the law where you and the server are, and by the provider's acceptable use policy. See seedbox privacy and security.
Some links on this page are affiliate links. If you sign up through them we may earn a commission, at no extra cost to you. It never changes which providers we recommend or how we rank them.
