homelab-blueprint/docker-compose-files/immich

Immich

Self-hosted photo and video management solution with high performance and visual aesthetics.

Prerequisites

• NVIDIA GPU: This template is configured for NVIDIA GPU acceleration (RTX A400).
• Proxmox LXC: Must be a Privileged LXC with NVIDIA drivers and Container Toolkit installed.
• Storage: NFS-backed storage for photo uploads is recommended.

Deployment

1. Update .env: Ensure the Immich variables are set in your root .env file. IMMICH_UPLOAD_LOCATION should point to your dedicated Immich share on the UNAS:

   IMMICH_UPLOAD_LOCATION=/mnt/unas/immich_data
   IMMICH_DATA_DIR=/opt/immich/appdata
   IMMICH_EXTERNAL_LIBRARY=/mnt/unas/immich_data/upload/upload
   IMMICH_MOBILE_SYNC_DIR=/mnt/unas/mobile_photos
   IMMICH_VERSION=release
   IMMICH_DB_PASSWORD=<your-secure-password>
   IMMICH_DB_USERNAME=immich
   IMMICH_DB_DATABASE_NAME=immich
   IMMICH_DB_HOSTNAME=database
   IMMICH_REDIS_HOSTNAME=redis
   

   Optional OIDC settings can also be supplied. Set OIDC_ENABLED=true explicitly when enabling OAuth; leaving it unset keeps Immich local-login only. Zitadel with PKCE does not require a client secret — leave OIDC_CLIENT_SECRET empty.

   OIDC_ENABLED=true
   OIDC_ISSUER=https://zitadel.example.com/
   OIDC_CLIENT_ID=<immich-client-id>
   OIDC_CLIENT_SECRET=
   OIDC_NAME=<button-label>
   OIDC_AUTO_REDIRECT=false
   OIDC_MOBILE_OVERRIDE_ENABLED=false
   OIDC_DISABLE_LOCAL_AUTH=false
   

   Once OIDC is confirmed working, set OIDC_DISABLE_LOCAL_AUTH=true and restart the server to enforce SSO-only login:

   sed -i 's/DISABLE_PASSWORD_LOGIN=false/DISABLE_PASSWORD_LOGIN=true/' /opt/immich/docker-compose.yml
   docker compose -f /opt/immich/docker-compose.yml up -d --no-deps --force-recreate immich-server
   

2. Create Directories: Create the local appdata directories and the shared media directory:

   # Local config/database storage (matches IMMICH_DATA_DIR)
   sudo mkdir -p /opt/immich/appdata/{postgres,model-cache}
   sudo chown -R 1000:1000 /opt/immich/appdata
   
   # Separate UNAS share for photos
   sudo mkdir -p /mnt/unas/immich_data
   sudo chown -R 1000:1000 /mnt/unas/immich_data
   

3. Mounting the Share: Ensure you have created a separate NFS share on your UNAS (e.g., immich_data) and mounted it on your Proxmox host at /mnt/unas/immich_data before starting the containers.

4. Render Compose File:

   ./substitute_env.sh docker-compose-files/immich/template.yaml docker-compose.immich.yml .env
   

5. Start Stack:

   docker compose -f docker-compose.immich.yml up -d
   

External Library Setup

After the stack is running, create an external library via the API to import photos from IMMICH_EXTERNAL_LIBRARY (mounted read-only at /mnt/external inside the container):

TOKEN=$(curl -s -X POST http://<immich-ip>:2283/api/auth/login \
  -H 'Content-Type: application/json' \
  -d '{"email":"<email>","password":"<password>"}' \
  | python3 -c 'import sys,json; print(json.load(sys.stdin)["accessToken"])')

USER_ID=$(curl -s http://<immich-ip>:2283/api/users/me \
  -H "Authorization: Bearer $TOKEN" \
  | python3 -c 'import sys,json; print(json.load(sys.stdin)["id"])')

LIB_ID=$(curl -s -X POST http://<immich-ip>:2283/api/libraries \
  -H 'Content-Type: application/json' \
  -H "Authorization: Bearer $TOKEN" \
  -d "{\"name\":\"NAS Photos\",\"ownerId\":\"$USER_ID\",\"importPaths\":[\"/mnt/external/<user-uuid>\"]}" \
  | python3 -c 'import sys,json; print(json.load(sys.stdin)["id"])')

curl -s -X POST http://<immich-ip>:2283/api/libraries/$LIB_ID/scan \
  -H "Authorization: Bearer $TOKEN"

The <user-uuid> is the Immich user's UUID — find it at Settings → Account in the Immich UI, or via:

curl -s http://<immich-ip>:2283/api/users/me -H "Authorization: Bearer $TOKEN" | python3 -c 'import sys,json; print(json.load(sys.stdin)["id"])'

Important

Immich rejects any external library import path that falls under /data/upload/ (the upload volume). The separate /mnt/external mount exists specifically to work around this restriction — the same photos are accessible at both paths inside the container, but only /mnt/external is accepted for external libraries.

Privileged Mode Rationale

Immich is deployed in a Privileged LXC to ensure:

• GPU Access: The immich-machine-learning container requires direct access to NVIDIA device nodes for CUDA compute.
• VRAM Management: CUDA memory allocation and GSP firmware communication are more stable in privileged mode.
• Storage Mapping: Simplified 1:1 UID/GID mapping for photo uploads on the UNAS NFS share.

GPU Verification

Both immich-server and immich-machine-learning run with runtime: nvidia:

• immich-server — uses NVENC for hardware-accelerated H.264 video transcoding (NVIDIA_DRIVER_CAPABILITIES=video,compute,utility)
• immich-machine-learning — uses the :release-cuda image with CUDAExecutionProvider for CLIP embeddings, face detection, and OCR

Verify GPU access after startup:

# Confirm NVIDIA device visible in server (needed for NVENC)
docker exec immich_server ls /dev/nvidia0

# Confirm CUDA ONNX provider active in ML container
docker exec immich_machine_learning python3 -c 'import onnxruntime; print(onnxruntime.get_available_providers())'
# Expected: ['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider']

After verifying, go to Administration → Video Transcoding in the Immich UI and confirm Hardware Acceleration is set to NVENC. If it shows QSV (from a prior config), change it to NVENC and re-run the Video Conversion job.