Skip to content

Step 4: Download Docker Templates

Quicklink to the Unraid terminal (paste into your browser while logged into Unraid):

http://<your-unraid-ip>/Terminal

Open the Unraid web interface and open the terminal (click the Terminal icon in the top right corner of the navigation bar). Then download the templates you need.

Note

wget downloads files from the web. The -P flag sets the download directory. Templates are saved to Unraid's Docker Manager template directory.

PostgreSQL

Choose between stability and latest features:

Option Image VectorChord pgvector Status
Stable ghcr.io/immich-app/postgres:18-vectorchord0.5.3-pgvector0.8.1 0.5.3 0.8.1 Tested by Immich team
Experimental tensorchord/vchord-postgres:pg18-v1.1.1 1.1.1 0.8.2 Latest, less tested

Note

You COULD use VectorChord 1.0.0+ (See #23845) or 1.1.1 (See this discussion). The 0.5.3 version is more stable and tested extensively with Immich. If you want to try the latest, use the experimental template.

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-postgres-official.xml

PostgreSQL database by VectorChord — EXPERIMENTAL (latest VectorChord 1.1.1, less tested with Immich):

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-vectorchord-db.xml

Valkey (cache/message broker):

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-valkey.xml

Machine Learning template (choose one based on your GPU):

CPU only — no GPU acceleration:

CPU inference is possible but will be much slower for tasks like face recognition and CLIP-based search. Only recommended if you have a powerful CPU and a small library. Will work but expect slower performance and high load on the CPU.

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-machine-learning.xml

NVIDIA CUDA:

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-machine-learning-cuda.xml

Intel OpenVINO:

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-machine-learning-openvino.xml

AMD ROCm:

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-machine-learning-rocm.xml

Server template (choose one based on your GPU):

These templates are for video transcoding. If you don't have a GPU or don't want to use it for transcoding, choose the CPU-only template.

CPU only — no GPU transcoding:

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-server.xml

Intel QSV / AMD VAAPI — uses /dev/dri:

You can also use your Intel iGPU ("inbuilt graphics card" in your CPU) for transcoding with Quick Sync Video (QSV) or your AMD GPU with VAAPI. Both use the same template since they both leverage /dev/dri for hardware acceleration.

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-server-qsv-vaapi.xml

NVIDIA NVENC — uses --runtime=nvidia:

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-server-nvenc.xml

PhotoMigrator (for Google Takeout migration):

wget -P /boot/config/plugins/dockerMan/templates-user/ https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/photomigrator.xml