immich-podman-quadlets/docs/install/podman-quadlet.md

# Podman deploy with quadlets You can deploy Immich on Podman using quadlets. Here are some sample rootless quadlet container files that can be placed in /etc/containers/systemd/users/${ID} where ID is the uid of whatever your rootless user is. With Immich version 1.106.1 the microservices container no longer has to be separate from the immich server container. That being said, in the case where you would prefer to offload to a separate GPU accelerated VM this is still the preferred method. The quadlets provided below give a sample of said scenario. Feel free to remove the Environment variable that declares the API include and exclude if you prefer just the one container to run both. Please note you'll need :z or :Z for selinux enabled hosts. If you are doing hardware acceleration with NVIDIA make sure you follow the immich documentation including the appropriate NVIDIA Container Device Interface (https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/cdi-support.html) and adding the device (AddDevice=nvidia.com/gpu=0). # Sample Quadlet Files It is recommended to run these rootless, and as such the ideal place is likely /etc/containers/systemd/users/${UID} ## Universal immich-database.container bash [Unit] Description=Immich Database Requires=immich-redis.service [Container] AutoUpdate=registry EnvironmentFile=${location_of_env_file} Image=registry.hub.docker.com/tensorchord/pgvecto-rs:pg16-v0.2.1 Label=registry Network=slirp4netns:port_handler=slirp4netns PublishPort=5432:5432 Volume=${host_database_directory}:/var/lib/postgresql/data:z Volume=${host_database_backup_directory}:/mnt/db_backup:z Volume=/etc/localtime:/etc/localtime:ro [Service] Restart=always [Install] WantedBy=multi-user.target default.target immich-redis.container bash [Unit] Description=Immich Redis [Container] AutoUpdate=registry Image=registry.hub.docker.com/library/redis:6.2-alpine@sha256:51d6c56749a4243096327e3fb964a48ed92254357108449cb6e23999c37773c5 Label=registry Network=slirp4netns:port_handler=slirp4netns PublishPort=6379:6379 Timezone=America/Montreal [Service] Restart=always [Install] WantedBy=multi-user.target default.target Alternatively you can use valkey bash [Unit] Description=Immich Keydb [Container] AutoUpdate=registry HealthCmd=redis-cli ping || exit 1 HealthInterval=1m HealthOnFailure=kill Image=docker.io/valkey/valkey:latest Label=registry Network=host PublishPort=6379:6379 Timezone=America/Montreal Volume=/mnt/data01/immich-valkey:/data:z [Service] Restart=always [Install] WantedBy=multi-user.target default.target immich-server.container bash [Unit] Description=Immich Server Requires=immich-redis.service immich-database.service [Container] AutoUpdate=registry EnvironmentFile=${location_of_env_file} Environment=IMMICH_WORKERS_INCLUDE='api' Image=ghcr.io/immich-app/immich-server:release Label=registry Network=slirp4netns:port_handler=slirp4netns PublishPort=3000:3000 PublishPort=3001:3001 Volume=${host_upload_directory}:/usr/src/app/upload Volume=/etc/localtime:/etc/localtime:ro [Service] Restart=always [Install] WantedBy=multi-user.target default.target ## No Hardware Acceleration immich-microservices.container bash [Unit] Description=Immich Microservices Requires=immich-redis.service immich-database.service [Container] AutoUpdate=registry EnvironmentFile=${location_of_env_file} Environment=IMMICH_WORKERS_EXCLUDE='api' Image=ghcr.io/immich-app/immich-server:release Label=registry Network=slirp4netns:port_handler=slirp4netns PublishPort=3002:3002 Volume=${host_upload_directory}:/usr/src/app/upload:z Volume=/etc/localtime:/etc/localtime:ro #Exec=start.sh microservices [Service] Restart=always [Install] WantedBy=multi-user.target default.target immich-ml.container bash [Unit] Description=Immich Machine Learning Requires=immich-redis.service immich-database.service [Container] AutoUpdate=registry EnvironmentFile=${location_of_env_file} Image=ghcr.io/immich-app/immich-machine-learning:release Label=registry Network=slirp4netns:port_handler=slirp4netns PublishPort=3003:3003 Volume=${cache_directory}:/cache:z Volume=/etc/localtime:/etc/localtime:ro [Service] Restart=always [Install] WantedBy=multi-user.target default.target ## Hardware Acceleration - NVIDIA CUDA immich-microservices.container bash [Unit] Description=Immich Microservices Requires=immich-redis.service immich-database.service [Container] AddDevice=/dev/dri:/dev/dri AddDevice=nvidia.com/gpu=0 AutoUpdate=registry EnvironmentFile=${location_of_env_file} Environment=IMMICH_WORKERS_EXCLUDE='api' Image=ghcr.io/immich-app/immich-server:${release} # This is for v118.0 and up Label=registry Network=slirp4netns:port_handler=slirp4netns PublishPort=3002:3002 Volume=/mnt/data01/uploads:/usr/src/app/upload:z Volume=/etc/localtime:/etc/localtime:ro Unmask=/dev/dri:/dev/dri [Service] Restart=always [Install] WantedBy=multi-user.target default.target immich-ml.container bash [Unit] Description=Immich Machine Learning Requires=mnt-data01.mount mnt-data01-uploads.mount After==mnt-data01.mount mnt-data01-uploads.mount [Container] AddDevice=/dev/dri:/dev/dri AddDevice=nvidia.com/gpu=0 # Make sure this matched your GPU ID AutoUpdate=registry EnvironmentFile=${location_of_env_file} Image=ghcr.io/immich-app/immich-machine-learning:${release} # This is for v118.0 and up Label=registry Network=slirp4netns:port_handler=slirp4netns #PublishPort=3003:3003 Volume=/mnt/data01/model-cache:/cache:z Volume=/etc/localtime:/etc/localtime:ro Unmask=/dev/dri:/dev/dri [Service] Restart=always [Install] WantedBy=multi-user.target default.target