Windows Install

🇬🇧 English · 🇫🇷 Français

Windows install

Hydra runs natively on Windows (64-bit). Tagged releases ship a self-contained zip — two .exe files, a config, nothing else to install. The web UI is embedded in the daemon.

Download

Grab the Windows zip from the Releases page:

• hydra-vX.Y.Z-windows-amd64.zip (Intel/AMD, 64-bit)

A .sha256 sits next to it if you want to verify the download.

The zip contains:

• hydra.exe — the daemon (web UI + API). This is the only one you launch.
• hydra-engine.exe — the Typhon BitTorrent engine. hydra.exe starts it automatically; you never run it yourself.
• default.toml.example — a starting configuration.
• README.md — a short version of this page.

Run

1. Unzip it anywhere, e.g. C:\Hydra.

2. Keep hydra.exe and hydra-engine.exe in the same folder — the daemon looks for the engine next to itself.

3. Copy default.toml.example to default.toml and edit at least:

   • data_dir — use a Windows path with doubled backslashes, e.g. data_dir = "C:\\Hydra\\data".
   • api_key — set your own.
   • the ports, if the defaults clash with something.

4. Start it from a terminal (PowerShell or cmd):

   hydra.exe --config default.toml
   

5. Open the web UI at http://127.0.0.1:8199 (or the api_port you set).

That's it. The daemon spawns hydra-engine.exe and talks to it over a TCP loopback socket on 127.0.0.1 — this is automatic on Windows and needs no configuration. (On Linux the two use a Unix domain socket instead; same wire protocol, nothing you interact with.)

Run in the background (as a service)

Running hydra.exe from a terminal keeps a console window open. To run it unattended and start it on boot, wrap it in a service manager such as NSSM or the built-in sc.exe. With NSSM:

nssm install Hydra "C:\Hydra\hydra.exe" --config "C:\Hydra\default.toml"
nssm start Hydra

If the memory watchdog restarts Hydra, the service manager brings it back up.

VPN

Hydra does not manage the VPN on Windows — and it doesn't need to. Use your VPN client's system-wide or per-app binding / kill-switch (Mullvad, AirVPN Eddie, Proton, PIA, etc.). All of Hydra's traffic then egresses through the tunnel like any other app.

The Linux-only per-socket routing (fwmark / SO_MARK, used for the multi-tunnel and gluetun setups) is disabled on Windows by design — see Gluetun (VPN) and Networking Modes for what those are and why they don't apply here.

Notes & limitations

• First run creates the data/resume folders under data_dir and generates an API key if you left it blank — check the console output for it.
• Windows Firewall may prompt to allow hydra.exe/hydra-engine.exe on first listen. Allow it on your private network for peers to reach you (or set up a port forward if you seed behind NAT).
• Heap profiling (jemalloc) is a Linux-only diagnostic and is absent from the Windows build; the system allocator is used instead. No functional difference for normal use.
• Everything else — categories, the qBittorrent shim, adding torrents, seeding existing data — works the same as on Linux. See the rest of the wiki.

Troubleshooting

• "engine endpoint not ready" on startup: hydra-engine.exe isn't next to hydra.exe, or an antivirus quarantined it. Keep both in the same folder and whitelist them.
• Web UI won't load: check the console for the actual api_port and that nothing else is bound to it; try http://127.0.0.1:<port> explicitly.
• Peers won't connect (inbound): allow the listen port through Windows Firewall and, behind a router, forward it. Outbound-only still seeds to peers that can reach you.

See Troubleshooting for the general guide.