Tdarr — Distributed Media Transcoding Automation

Introduction

Tdarr is a distributed transcoding application that processes your media library in bulk. It scans video and audio files, applies user-defined rules through a flow-based plugin system, and distributes transcode jobs across multiple worker nodes using FFmpeg or HandBrake.

What Tdarr Does

• Scans media libraries for files that do not match your desired codec, container, or bitrate
• Distributes transcode and health-check jobs across CPU and GPU worker nodes
• Provides a visual flow editor for building complex conditional processing pipelines
• Reports library-wide codec statistics and storage savings after transcoding
• Runs automated health checks to detect corrupt or truncated media files

Architecture Overview

Tdarr uses a server-node model. The server coordinates the job queue and hosts the web UI on port 8265. Worker nodes connect to the server and pull transcode or health-check tasks. Each node can run multiple CPU and GPU workers in parallel. Processing uses FFmpeg or HandBrake under the hood, selected per flow step. The internal node option runs a worker inside the server container for single-machine setups.

Self-Hosting & Configuration

• Deploy the server with Docker and optionally add remote worker nodes on other machines
• Mount your media library and a temporary transcode directory into the container
• Create libraries in the web UI and point them at your media folders
• Build processing flows using the visual drag-and-drop editor or use community presets
• Enable GPU transcoding by passing through NVIDIA or Intel Quick Sync devices

Key Features

• Distributed architecture scales from one machine to a cluster of worker nodes
• Visual flow editor for building conditional transcode and remux pipelines
• GPU-accelerated transcoding with NVENC, QSV, and VAAPI support
• Library analytics showing codec breakdown, resolution distribution, and space savings
• Community plugin repository with hundreds of ready-made processing flows

Comparison with Similar Tools

• HandBrake — standalone GUI/CLI transcoder, no library scanning or distributed workers
• FFmpeg — command-line only, no queue management or web UI
• Unmanic — similar library optimizer, single-node only, smaller community
• FileFlows — flow-based file processing, supports more file types beyond media
• Bazarr — subtitle manager for media, does not handle transcoding

FAQ

Q: Does Tdarr re-encode files automatically? A: Only files that match your flow conditions are processed. You define which codecs, containers, or quality levels trigger a transcode.

Q: Can I use GPU transcoding? A: Yes. Pass through your NVIDIA GPU or Intel iGPU to the container and configure GPU workers in the node settings. NVENC and Quick Sync are both supported.

Q: Will Tdarr damage my original files? A: Tdarr transcodes to a temporary directory first and only replaces the original after a successful health check. You can also enable a cache or backup directory for extra safety.

Q: How many nodes can I run? A: There is no hard limit. Each node registers with the server and pulls jobs independently. Add more nodes to increase throughput.

Sources

• https://github.com/HaveAGitGat/Tdarr
• https://docs.tdarr.io/