FileFlows

An automated media processing and transcoding pipeline server.

FileFlows is a self-hosted media file processing tool that automates transcoding, remuxing, and moving media files through configurable processing pipelines. It integrates with GPU hardware for accelerated transcoding. Self-hosting gives you full control over your media library processing without relying on cloud services.

Alternatives considered

Self Hosted

Tool	Open Source	Full Features	Notes
Tdarr	no	Limited	Popular transcoding pipeline, closed core
Unmanic	Yes	Yes	Plugin-based automated transcoding
HandBrake	Yes	Yes	Manual only, no automation

Installation

Architecture

• Deployment: Single fileflows deployment in the fileflows namespace
• Image: revenz/fileflows:stable (digest-pinned to sha256:2558e8…)
• Storage: Longhorn PVC (fileflows-data, 1Gi, longhorn-encrypted); NFS volume from TrueNAS read-only for media input
• Resources: Intel GPU resource limit (gpu.intel.com/i915: 1) for hardware-accelerated transcoding
• Networking: ClusterIP service on port 5000, HTTPRoute via internal gateway

Security

No container-level securityContext configured. Longhorn PVC encrypted at rest via SOPS-managed key.

Updates

Managed by Renovate. Image is digest-pinned (tracks stable tag).

Data Management

• PVC: fileflows-data (1Gi, longhorn-encrypted) for app config and processing state
• NFS: TrueNAS NFS mount for media input (read-only)
• Backups: k8up daily schedule (Hetzner + Backblaze) for fileflows-data, which holds the API-managed flows that exist nowhere else; fileflows-logs is excluded (k8up.io/backup: "false")

User Management

No OIDC or external authentication configured. Access via internal-only HTTPRoute.

Configuration Management

• Longhorn encryption key from SOPS-encrypted secret
• Intel GPU scheduling configured via resource limits in the Deployment

Usage

Define processing flows (pipelines) in the web UI to automatically transcode or remux incoming media files. Flows can branch on file properties (codec, resolution, container). The Intel GPU integration enables hardware-accelerated H.264/H.265 encoding for faster processing.

Cluster-specific deviations from the above live in the per-cluster README — see k8s/apps/fileflows/talos/README.md.

Cluster Deployment

App URLs

• fileflows.int.kueber.eu

Depends on

• Envoy Gateway
• k8up backups
• Longhorn

FileFlows — Talos cluster

Deviations from defaults

• fileflows-logs is not backed up (k8up.io/backup: "false" in backups/patch-pvcs.yaml): rotating log files, nothing to restore. fileflows-data (the API-managed flows, not in git) is.

Kubernetes Metadata

• Image: revenz/fileflows:26.10@sha256:b35d2ed5e725ec74b4ead73eaae062bd9fa57837d2bce7fda85192b91af5d562

Rendered manifests (kustomize build)

apiVersion: apps/v1
kind: Deployment
metadata:
  annotations:
    kustomize.toolkit.fluxcd.io/force: enabled
  labels:
    app.kubernetes.io/name: fileflows
  name: fileflows
  namespace: fileflows
spec:
  replicas: 1
  revisionHistoryLimit: 2
  selector:
    matchLabels:
      app.kubernetes.io/name: fileflows
  strategy:
    rollingUpdate: null
    type: Recreate
  template:
    metadata:
      labels:
        app.kubernetes.io/name: fileflows
    spec:
      automountServiceAccountToken: false
      containers:
        - image: revenz/fileflows:26.10@sha256:b35d2ed5e725ec74b4ead73eaae062bd9fa57837d2bce7fda85192b91af5d562
          livenessProbe:
            failureThreshold: 5
            httpGet:
              path: /
              port: http
            initialDelaySeconds: 5
            periodSeconds: 10
            successThreshold: 1
            timeoutSeconds: 5
          name: fileflows
          ports:
            - containerPort: 5000
              name: http
              protocol: TCP
          readinessProbe:
            failureThreshold: 5
            httpGet:
              path: /
              port: http
            initialDelaySeconds: 5
            periodSeconds: 10
            successThreshold: 1
            timeoutSeconds: 5
          resources:
            limits:
              gpu.intel.com/i915: 1
            requests:
              cpu: 100m
              gpu.intel.com/i915: 1
              memory: 768Mi
          volumeMounts:
            - mountPath: /media/stash
              name: stash
            - mountPath: /app/Data
              name: fileflows-data
            - mountPath: /app/Logs
              name: fileflows-logs
      priorityClassName: tier-3
      terminationGracePeriodSeconds: 60
      volumes:
        - name: stash
          persistentVolumeClaim:
            claimName: fileflows-truenas-nfs-stash
        - name: fileflows-data
          persistentVolumeClaim:
            claimName: fileflows-data
        - name: fileflows-logs
          persistentVolumeClaim:
            claimName: fileflows-logs