Jellyfin with Docker Compose
Jellyfin is a Free Software Media System that puts you in control of managing and streaming your media, from a dedicated server to end-user devices via multiple apps. In the project's words: Jellyfin README checked 2026-09-23
Replaces: Emby and Plex. The project calls itself "an alternative to the proprietary Emby and Plex". Jellyfin README checked 2026-09-23
Runs on: amd64, arm64 from each image's registry manifest, checked 2026-09-25; a CPU counts only if every image has a build for it. Image by image
Add Jellyfin to the builder to use your own domain and combine it with other apps in one file.
What Jellyfin needs
Only figures Jellyfin's own documentation publishes, each linked to the page it is on. Where the project gives no number, this page does not invent one.
| Resource | Minimum | Recommended |
|---|---|---|
| RAM | 4 GB | 8 GB |
| 8 GB is the recommendation for an average deployment; 4 GB "may be enough" for a Linux server without a desktop. Add more on Windows 11. Source: Jellyfin docs: Hardware Selection (System Memory) checked 2026-09-23 | ||
| CPU | Any modern CPU with 4 threads | Intel Core i5-11400, Pentium Gold G7400, N100 or Apple M series |
| Both figures assume hardware-accelerated transcoding (an Intel iGPU or a dedicated GPU). Running without a GPU is "NOT recommended": depending on the configuration, the docs say, a Ryzen 9 5950X may not handle even a single video stream. x86 CPUs need SSE4.1 from Jellyfin 10.11. Source: Jellyfin docs: Hardware Selection checked 2026-09-23 | ||
| Disk | not stated | 100 GB SSD |
| For the OS, Jellyfin's own files and the transcoding cache; the media library is extra and may sit on any storage faster than your media's bitrate, network shares included. Source: Jellyfin docs: Hardware Selection (Shared Items, Storage) checked 2026-09-23 | ||
Which hardware Jellyfin has images for
Read from each image's manifest list in its registry (anonymous, read-only) when this page was built. An app counts for a CPU only if every one of its images lists it.
- x86-64 PC or NAS (
linux/amd64): every image has a build. - Raspberry Pi 4/5, 64-bit OS (
linux/arm64): every image has a build. - Raspberry Pi 3 or Zero 2 W, 32-bit OS (
linux/arm/v7): no image forjellyfin. No image for this hardware; the builder leaves the app out.
| Service | Image | Platforms in the manifest | Digest | Registry | Checked |
|---|---|---|---|---|---|
jellyfin | jellyfin/jellyfin | amd64, arm64 | sha256:78d3ea1207d1322471fcac39a614f004f2ccf7e878f95ab2977d752f07e4dd7e | registry-1.docker.io | 2026-09-25 |
Your hardware
Jellyfin converts video the player cannot play as it streams. Without a GPU passed into the container it does that on the processor. Pick what the server has: the compose file below then carries the lines Jellyfin's hardware acceleration pages give for it, each quoted from those pages, and the notes say what to switch on in Jellyfin afterwards.
LAN discovery
Jellyfin's networking page: “Additionally, in local networks, Jellyfin offers various Auto-Discovery services. These will not work outside your local subnet.” Jellyfin docs: Networking (port bindings, auto-discovery, firewall) checked 2026-10-09 Off, the compose file below leaves it out, so each Jellyfin app must be given the server's address by hand. On, it publishes 7359/udp beside Caddy, labelled with its purpose. The web port, 8096, stays behind Caddy either way.
What on publishes
7359:7359/udp(client discovery): “Allows clients to discover Jellyfin on the local network. A broadcast message to this port will return detailed information about your server that includes name, ip-address and ID.” Jellyfin docs: Networking (port bindings, auto-discovery, firewall) checked 2026-10-09- The mapping, as the Docker docs write it: “- 7359:7359/udp” Jellyfin docs: Container install (Docker Compose) checked 2026-10-09
- The address discovery hands out: the file already sets JELLYFIN_PublishedServerUrl to the HTTPS address behind Caddy, the variable the compose example comments as “# Optional - alternative address used for autodiscovery” Jellyfin docs: Container install (Docker Compose) checked 2026-10-09
- DLNA is a different thing, and this switch leaves it out; the container page says: “Using host networking (--net=host) is optional but required in order to use DLNA.” Jellyfin docs: Container install (Docker Compose) checked 2026-10-09
Firewall: if the server runs one, Jellyfin's container page says in its firewalld steps: “Additionally port 7359 UDP needs to be opened for auto discovery.” Jellyfin docs: Container install (Docker Compose) checked 2026-10-09
Do not forward 7359/udp from your router, and do not turn this on for a server with a public IP address (a VPS). Jellyfin: “Automatic discovery only works locally and should not be exposed externally.” Jellyfin docs: Networking (port bindings, auto-discovery, firewall) checked 2026-10-09
docker-compose.yml
None (CPU)
This is exactly what the builder writes for Jellyfin alone on example.org: the official definition with its published ports removed and Caddy added in front.
# docker-compose.yml built at dankhost.com for: Jellyfin # Each app block is its official compose definition. Only the names were prefixed with # the app slug (services, containers, volumes, bind folders, .env variables) so any set of # apps can share one file; other differences are listed under "Changed:" in each block. # Only Caddy publishes ports. Secrets live in .env next to this file. services: # ---- Caddy: the one reverse proxy, HTTPS certificates from Let's Encrypt ---- # Source: https://caddyserver.com/docs/running (checked 2026-09-23) # Source: https://github.com/dani-garcia/vaultwarden/wiki/Using-Docker-Compose (checked 2026-09-23) # Caddyfile goes in ./caddy/Caddyfile. Fixed address 172.30.0.2 so apps can trust it exactly. caddy: image: "caddy:2" container_name: "caddy" restart: "unless-stopped" ports: - "80:80" - "443:443" - "443:443/udp" volumes: - "./caddy:/etc/caddy" - "caddy-data:/data" - "caddy-config:/config" networks: proxy: ipv4_address: "172.30.0.2" # ---- Jellyfin (media.example.org) ---- # Source: https://jellyfin.org/docs/general/installation/container/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/caddy/ (checked 2026-09-23) # Changed: Removed the documented '8096:8096/tcp' and '7359:7359/udp' ports. 7359/udp is only for client auto-discovery on your LAN, so apps must be pointed at the address by hand. # The documented /path/to/config, /path/to/cache and /path/to/media placeholders become folders next to the compose file; the optional second media folder, fonts folder and 'user:' line are left out. jellyfin: image: "jellyfin/jellyfin" container_name: "jellyfin" volumes: - "./jellyfin/config:/config" - "./jellyfin/cache:/cache" - "./jellyfin/media:/media" restart: "unless-stopped" environment: JELLYFIN_PublishedServerUrl: "https://media.example.org" networks: proxy: {} networks: # Pinned so the trusted-proxy address given to each app (172.30.0.2) is exact. # If "docker compose up" reports "Pool overlaps", change 172.30.0 here, in Caddy's # ipv4_address and in every trusted-proxy value to a free 172.x.0 range. proxy: ipam: config: - subnet: "172.30.0.0/24" ip_range: "172.30.0.128/25" gateway: "172.30.0.1" volumes: caddy-data: caddy-config:
None (CPU), LAN discovery on
The None (CPU) file with 7359:7359/udp published beside Caddy, each labelled # PORT OUTSIDE THE PROXY with its purpose (see LAN discovery). Everything else is the same file as with LAN discovery off.
# docker-compose.yml built at dankhost.com for: Jellyfin # Each app block is its official compose definition. Only the names were prefixed with # the app slug (services, containers, volumes, bind folders, .env variables) so any set of # apps can share one file; other differences are listed under "Changed:" in each block. # Caddy publishes 80 and 443. The only other published ports, each labelled where # it is written: 7359/udp (Jellyfin: client discovery). # Secrets live in .env next to this file. services: # ---- Caddy: the one reverse proxy, HTTPS certificates from Let's Encrypt ---- # Source: https://caddyserver.com/docs/running (checked 2026-09-23) # Source: https://github.com/dani-garcia/vaultwarden/wiki/Using-Docker-Compose (checked 2026-09-23) # Caddyfile goes in ./caddy/Caddyfile. Fixed address 172.30.0.2 so apps can trust it exactly. caddy: image: "caddy:2" container_name: "caddy" restart: "unless-stopped" ports: - "80:80" - "443:443" - "443:443/udp" volumes: - "./caddy:/etc/caddy" - "caddy-data:/data" - "caddy-config:/config" networks: proxy: ipv4_address: "172.30.0.2" # ---- Jellyfin (media.example.org) ---- # Source: https://jellyfin.org/docs/general/installation/container/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/caddy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/ (checked 2026-10-09) # Changed: Removed the documented '8096:8096/tcp' port; Caddy reaches jellyfin:8096 on the proxy network. LAN discovery is on: the documented '7359:7359/udp', client auto-discovery on your LAN, is published beside Caddy. # The documented /path/to/config, /path/to/cache and /path/to/media placeholders become folders next to the compose file; the optional second media folder, fonts folder and 'user:' line are left out. # PORT OUTSIDE THE PROXY 7359/udp: “Allows clients to discover Jellyfin on the local network.” (Jellyfin's networking docs). For your home network only; do not forward it on your router. jellyfin: image: "jellyfin/jellyfin" container_name: "jellyfin" volumes: - "./jellyfin/config:/config" - "./jellyfin/cache:/cache" - "./jellyfin/media:/media" restart: "unless-stopped" environment: JELLYFIN_PublishedServerUrl: "https://media.example.org" ports: - "7359:7359/udp" # client discovery networks: proxy: {} networks: # Pinned so the trusted-proxy address given to each app (172.30.0.2) is exact. # If "docker compose up" reports "Pool overlaps", change 172.30.0 here, in Caddy's # ipv4_address and in every trusted-proxy value to a free 172.x.0 range. proxy: ipam: config: - subnet: "172.30.0.0/24" ip_range: "172.30.0.128/25" gateway: "172.30.0.1" volumes: caddy-data: caddy-config:
With Intel / AMD
Intel integrated graphics or an Intel Arc card (QSV or VA-API), or an AMD card or APU (VA-API). Jellyfin's Intel and AMD pages give the same two compose lines. For Intel or AMD graphics: group_add '122' and devices: /dev/dri/renderD128:/dev/dri/renderD128 from the Official Docker section of Jellyfin's Intel and AMD hardware acceleration pages. 122 is the docs' example render group id: replace it with what getent group render | cut -d: -f3 prints on your host. The docs' example also has user: 1000:1000 and the host network; this file keeps the image's default user and the proxy network. Everything else is the None file, as the builder writes it for Jellyfin alone on example.org.
# docker-compose.yml built at dankhost.com for: Jellyfin # Each app block is its official compose definition. Only the names were prefixed with # the app slug (services, containers, volumes, bind folders, .env variables) so any set of # apps can share one file; other differences are listed under "Changed:" in each block. # Only Caddy publishes ports. Secrets live in .env next to this file. services: # ---- Caddy: the one reverse proxy, HTTPS certificates from Let's Encrypt ---- # Source: https://caddyserver.com/docs/running (checked 2026-09-23) # Source: https://github.com/dani-garcia/vaultwarden/wiki/Using-Docker-Compose (checked 2026-09-23) # Caddyfile goes in ./caddy/Caddyfile. Fixed address 172.30.0.2 so apps can trust it exactly. caddy: image: "caddy:2" container_name: "caddy" restart: "unless-stopped" ports: - "80:80" - "443:443" - "443:443/udp" volumes: - "./caddy:/etc/caddy" - "caddy-data:/data" - "caddy-config:/config" networks: proxy: ipv4_address: "172.30.0.2" # ---- Jellyfin (media.example.org) ---- # Source: https://jellyfin.org/docs/general/installation/container/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/caddy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/transcoding/hardware-acceleration/intel/ (checked 2026-10-09) # Source: https://jellyfin.org/docs/general/post-install/transcoding/hardware-acceleration/amd/ (checked 2026-10-09) # Changed: Removed the documented '8096:8096/tcp' and '7359:7359/udp' ports. 7359/udp is only for client auto-discovery on your LAN, so apps must be pointed at the address by hand. # The documented /path/to/config, /path/to/cache and /path/to/media placeholders become folders next to the compose file; the optional second media folder, fonts folder and 'user:' line are left out. # For Intel or AMD graphics: group_add '122' and devices: /dev/dri/renderD128:/dev/dri/renderD128 from the Official Docker section of Jellyfin's Intel and AMD hardware acceleration pages. 122 is the docs' example render group id: replace it with what getent group render | cut -d: -f3 prints on your host. The docs' example also has user: 1000:1000 and the host network; this file keeps the image's default user and the proxy network. jellyfin: image: "jellyfin/jellyfin" container_name: "jellyfin" group_add: - "122" devices: - "/dev/dri/renderD128:/dev/dri/renderD128" volumes: - "./jellyfin/config:/config" - "./jellyfin/cache:/cache" - "./jellyfin/media:/media" restart: "unless-stopped" environment: JELLYFIN_PublishedServerUrl: "https://media.example.org" networks: proxy: {} networks: # Pinned so the trusted-proxy address given to each app (172.30.0.2) is exact. # If "docker compose up" reports "Pool overlaps", change 172.30.0 here, in Caddy's # ipv4_address and in every trusted-proxy value to a free 172.x.0 range. proxy: ipam: config: - subnet: "172.30.0.0/24" ip_range: "172.30.0.128/25" gateway: "172.30.0.1" volumes: caddy-data: caddy-config:
Then, in Jellyfin: “Enable QSV or VA-API in Jellyfin and uncheck the unsupported codecs.” Jellyfin docs: Hardware acceleration on Intel GPU (Official Docker) checked 2026-10-09
jellyfin/jellyfin: “image: jellyfin/jellyfin” Jellyfin docs: Hardware acceleration on Intel GPU (Official Docker) checked 2026-10-09122: Jellyfin docs: Hardware acceleration on Intel GPU (Official Docker) checked 2026-10-09group_add: - '122' # Change this to match your "render" host group id and remove this comment
/dev/dri/renderD128:/dev/dri/renderD128: Jellyfin docs: Hardware acceleration on Intel GPU (Official Docker) checked 2026-10-09devices: - /dev/dri/renderD128:/dev/dri/renderD128
- Find your render group id on the host with: “getent group render | cut -d: -f3” Jellyfin docs: Hardware acceleration on Intel GPU (Official Docker) checked 2026-10-09
- The group name differs on some systems: “On some releases, the group may be video or input instead of render.” Jellyfin docs: Hardware acceleration on Intel GPU (Official Docker) checked 2026-10-09
- The AMD page gives the same compose lines and also has you look up the video group id: Jellyfin docs: Hardware acceleration on AMD GPU (Official Docker) checked 2026-10-09
getent group render | cut -d: -f3 getent group video | cut -d: -f3
- renderD128 is the first GPU; with two: “If you wish to use the second GPU on your system, change renderD128 to renderD129.” Jellyfin docs: Hardware acceleration on Intel GPU (Official Docker) checked 2026-10-09
- On AMD, the method to pick in Jellyfin is VA-API: “Enable VA-API in Jellyfin and uncheck the unsupported codecs.” Jellyfin docs: Hardware acceleration on AMD GPU (Official Docker) checked 2026-10-09
With Intel / AMD, LAN discovery on
The Intel / AMD file with 7359:7359/udp published beside Caddy, each labelled # PORT OUTSIDE THE PROXY with its purpose (see LAN discovery). Everything else is the same file as with LAN discovery off.
# docker-compose.yml built at dankhost.com for: Jellyfin # Each app block is its official compose definition. Only the names were prefixed with # the app slug (services, containers, volumes, bind folders, .env variables) so any set of # apps can share one file; other differences are listed under "Changed:" in each block. # Caddy publishes 80 and 443. The only other published ports, each labelled where # it is written: 7359/udp (Jellyfin: client discovery). # Secrets live in .env next to this file. services: # ---- Caddy: the one reverse proxy, HTTPS certificates from Let's Encrypt ---- # Source: https://caddyserver.com/docs/running (checked 2026-09-23) # Source: https://github.com/dani-garcia/vaultwarden/wiki/Using-Docker-Compose (checked 2026-09-23) # Caddyfile goes in ./caddy/Caddyfile. Fixed address 172.30.0.2 so apps can trust it exactly. caddy: image: "caddy:2" container_name: "caddy" restart: "unless-stopped" ports: - "80:80" - "443:443" - "443:443/udp" volumes: - "./caddy:/etc/caddy" - "caddy-data:/data" - "caddy-config:/config" networks: proxy: ipv4_address: "172.30.0.2" # ---- Jellyfin (media.example.org) ---- # Source: https://jellyfin.org/docs/general/installation/container/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/caddy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/transcoding/hardware-acceleration/intel/ (checked 2026-10-09) # Source: https://jellyfin.org/docs/general/post-install/transcoding/hardware-acceleration/amd/ (checked 2026-10-09) # Source: https://jellyfin.org/docs/general/post-install/networking/ (checked 2026-10-09) # Changed: Removed the documented '8096:8096/tcp' port; Caddy reaches jellyfin:8096 on the proxy network. LAN discovery is on: the documented '7359:7359/udp', client auto-discovery on your LAN, is published beside Caddy. # The documented /path/to/config, /path/to/cache and /path/to/media placeholders become folders next to the compose file; the optional second media folder, fonts folder and 'user:' line are left out. # For Intel or AMD graphics: group_add '122' and devices: /dev/dri/renderD128:/dev/dri/renderD128 from the Official Docker section of Jellyfin's Intel and AMD hardware acceleration pages. 122 is the docs' example render group id: replace it with what getent group render | cut -d: -f3 prints on your host. The docs' example also has user: 1000:1000 and the host network; this file keeps the image's default user and the proxy network. # PORT OUTSIDE THE PROXY 7359/udp: “Allows clients to discover Jellyfin on the local network.” (Jellyfin's networking docs). For your home network only; do not forward it on your router. jellyfin: image: "jellyfin/jellyfin" container_name: "jellyfin" group_add: - "122" devices: - "/dev/dri/renderD128:/dev/dri/renderD128" volumes: - "./jellyfin/config:/config" - "./jellyfin/cache:/cache" - "./jellyfin/media:/media" restart: "unless-stopped" environment: JELLYFIN_PublishedServerUrl: "https://media.example.org" ports: - "7359:7359/udp" # client discovery networks: proxy: {} networks: # Pinned so the trusted-proxy address given to each app (172.30.0.2) is exact. # If "docker compose up" reports "Pool overlaps", change 172.30.0 here, in Caddy's # ipv4_address and in every trusted-proxy value to a free 172.x.0 range. proxy: ipam: config: - subnet: "172.30.0.0/24" ip_range: "172.30.0.128/25" gateway: "172.30.0.1" volumes: caddy-data: caddy-config:
With NVIDIA
An NVIDIA card with the NVIDIA driver and the NVIDIA Container Toolkit installed on the host (NVENC/NVDEC). For an NVIDIA card: runtime: nvidia and the deploy: GPU reservation (driver: nvidia, count: all, capabilities: [gpu]) from the Official Docker section of Jellyfin's NVIDIA hardware acceleration page. The docs' example also has user: 1000:1000 and the host network; this file keeps the image's default user and the proxy network. Everything else is the None file, as the builder writes it for Jellyfin alone on example.org.
# docker-compose.yml built at dankhost.com for: Jellyfin # Each app block is its official compose definition. Only the names were prefixed with # the app slug (services, containers, volumes, bind folders, .env variables) so any set of # apps can share one file; other differences are listed under "Changed:" in each block. # Only Caddy publishes ports. Secrets live in .env next to this file. services: # ---- Caddy: the one reverse proxy, HTTPS certificates from Let's Encrypt ---- # Source: https://caddyserver.com/docs/running (checked 2026-09-23) # Source: https://github.com/dani-garcia/vaultwarden/wiki/Using-Docker-Compose (checked 2026-09-23) # Caddyfile goes in ./caddy/Caddyfile. Fixed address 172.30.0.2 so apps can trust it exactly. caddy: image: "caddy:2" container_name: "caddy" restart: "unless-stopped" ports: - "80:80" - "443:443" - "443:443/udp" volumes: - "./caddy:/etc/caddy" - "caddy-data:/data" - "caddy-config:/config" networks: proxy: ipv4_address: "172.30.0.2" # ---- Jellyfin (media.example.org) ---- # Source: https://jellyfin.org/docs/general/installation/container/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/caddy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/transcoding/hardware-acceleration/intel/ (checked 2026-10-09) # Source: https://jellyfin.org/docs/general/post-install/transcoding/hardware-acceleration/nvidia/ (checked 2026-10-09) # Changed: Removed the documented '8096:8096/tcp' and '7359:7359/udp' ports. 7359/udp is only for client auto-discovery on your LAN, so apps must be pointed at the address by hand. # The documented /path/to/config, /path/to/cache and /path/to/media placeholders become folders next to the compose file; the optional second media folder, fonts folder and 'user:' line are left out. # For an NVIDIA card: runtime: nvidia and the deploy: GPU reservation (driver: nvidia, count: all, capabilities: [gpu]) from the Official Docker section of Jellyfin's NVIDIA hardware acceleration page. The docs' example also has user: 1000:1000 and the host network; this file keeps the image's default user and the proxy network. jellyfin: image: "jellyfin/jellyfin" container_name: "jellyfin" volumes: - "./jellyfin/config:/config" - "./jellyfin/cache:/cache" - "./jellyfin/media:/media" restart: "unless-stopped" environment: JELLYFIN_PublishedServerUrl: "https://media.example.org" runtime: "nvidia" deploy: resources: reservations: devices: - driver: "nvidia" count: "all" capabilities: - "gpu" networks: proxy: {} networks: # Pinned so the trusted-proxy address given to each app (172.30.0.2) is exact. # If "docker compose up" reports "Pool overlaps", change 172.30.0 here, in Caddy's # ipv4_address and in every trusted-proxy value to a free 172.x.0 range. proxy: ipam: config: - subnet: "172.30.0.0/24" ip_range: "172.30.0.128/25" gateway: "172.30.0.1" volumes: caddy-data: caddy-config:
Then, in Jellyfin: “Enable NVENC in Jellyfin and uncheck the unsupported codecs.” Jellyfin docs: Hardware acceleration on NVIDIA GPU (Official Docker) checked 2026-10-09
jellyfin/jellyfin: “image: jellyfin/jellyfin” Jellyfin docs: Hardware acceleration on Intel GPU (Official Docker) checked 2026-10-09nvidia: “ runtime: nvidia” Jellyfin docs: Hardware acceleration on NVIDIA GPU (Official Docker) checked 2026-10-09driver: nvidia,count: all,capabilities: [gpu]: Jellyfin docs: Hardware acceleration on NVIDIA GPU (Official Docker) checked 2026-10-09- driver: nvidia count: all capabilities: [gpu]
- Install both on the host before starting the file; without the toolkit, Docker refuses runtime: nvidia. “You have to install the NVIDIA driver and NVIDIA Container Toolkit on the host system to allow Docker access to your GPU.” Jellyfin docs: Hardware acceleration on NVIDIA GPU (Official Docker) checked 2026-10-09
- If transcoding fails with this error, the NVIDIA page lists six /dev/nvidia* devices to add as well (/dev/nvidia-caps, /dev/nvidia0, /dev/nvidiactl, /dev/nvidia-modeset, /dev/nvidia-uvm, /dev/nvidia-uvm-tools): “If you encounter the upstream issue CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected, pass these extra devices to the Docker:” Jellyfin docs: Hardware acceleration on NVIDIA GPU (Official Docker) checked 2026-10-09
With NVIDIA, LAN discovery on
The NVIDIA file with 7359:7359/udp published beside Caddy, each labelled # PORT OUTSIDE THE PROXY with its purpose (see LAN discovery). Everything else is the same file as with LAN discovery off.
# docker-compose.yml built at dankhost.com for: Jellyfin # Each app block is its official compose definition. Only the names were prefixed with # the app slug (services, containers, volumes, bind folders, .env variables) so any set of # apps can share one file; other differences are listed under "Changed:" in each block. # Caddy publishes 80 and 443. The only other published ports, each labelled where # it is written: 7359/udp (Jellyfin: client discovery). # Secrets live in .env next to this file. services: # ---- Caddy: the one reverse proxy, HTTPS certificates from Let's Encrypt ---- # Source: https://caddyserver.com/docs/running (checked 2026-09-23) # Source: https://github.com/dani-garcia/vaultwarden/wiki/Using-Docker-Compose (checked 2026-09-23) # Caddyfile goes in ./caddy/Caddyfile. Fixed address 172.30.0.2 so apps can trust it exactly. caddy: image: "caddy:2" container_name: "caddy" restart: "unless-stopped" ports: - "80:80" - "443:443" - "443:443/udp" volumes: - "./caddy:/etc/caddy" - "caddy-data:/data" - "caddy-config:/config" networks: proxy: ipv4_address: "172.30.0.2" # ---- Jellyfin (media.example.org) ---- # Source: https://jellyfin.org/docs/general/installation/container/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/networking/reverse-proxy/caddy/ (checked 2026-09-23) # Source: https://jellyfin.org/docs/general/post-install/transcoding/hardware-acceleration/intel/ (checked 2026-10-09) # Source: https://jellyfin.org/docs/general/post-install/transcoding/hardware-acceleration/nvidia/ (checked 2026-10-09) # Source: https://jellyfin.org/docs/general/post-install/networking/ (checked 2026-10-09) # Changed: Removed the documented '8096:8096/tcp' port; Caddy reaches jellyfin:8096 on the proxy network. LAN discovery is on: the documented '7359:7359/udp', client auto-discovery on your LAN, is published beside Caddy. # The documented /path/to/config, /path/to/cache and /path/to/media placeholders become folders next to the compose file; the optional second media folder, fonts folder and 'user:' line are left out. # For an NVIDIA card: runtime: nvidia and the deploy: GPU reservation (driver: nvidia, count: all, capabilities: [gpu]) from the Official Docker section of Jellyfin's NVIDIA hardware acceleration page. The docs' example also has user: 1000:1000 and the host network; this file keeps the image's default user and the proxy network. # PORT OUTSIDE THE PROXY 7359/udp: “Allows clients to discover Jellyfin on the local network.” (Jellyfin's networking docs). For your home network only; do not forward it on your router. jellyfin: image: "jellyfin/jellyfin" container_name: "jellyfin" volumes: - "./jellyfin/config:/config" - "./jellyfin/cache:/cache" - "./jellyfin/media:/media" restart: "unless-stopped" environment: JELLYFIN_PublishedServerUrl: "https://media.example.org" runtime: "nvidia" deploy: resources: reservations: devices: - driver: "nvidia" count: "all" capabilities: - "gpu" ports: - "7359:7359/udp" # client discovery networks: proxy: {} networks: # Pinned so the trusted-proxy address given to each app (172.30.0.2) is exact. # If "docker compose up" reports "Pool overlaps", change 172.30.0 here, in Caddy's # ipv4_address and in every trusted-proxy value to a free 172.x.0 range. proxy: ipam: config: - subnet: "172.30.0.0/24" ip_range: "172.30.0.128/25" gateway: "172.30.0.1" volumes: caddy-data: caddy-config:
Caddyfile save as caddy/Caddyfile
Caddy gets and renews the HTTPS certificate for media.example.org by itself once the DNS record points at your server and ports 80 and 443 reach it.
# Caddyfile built at dankhost.com. Save it as ./caddy/Caddyfile next to docker-compose.yml.
# Caddy gets and renews a Let's Encrypt certificate for each name below by itself,
# once the DNS records point at this server and ports 80 and 443 reach it.
# Jellyfin
media.example.org {
reverse_proxy jellyfin:8096
}
.env
Jellyfin needs no secret here; the builder adds nothing you have to fill in by hand.
# .env for Jellyfin, from dankhost.com/app/jellyfin/. Nothing here is secret. # Keep this file private (chmod 600). # None of the chosen apps needs a variable here.
Changes from the official file
Besides prefixing every service, container, volume, folder and variable with jellyfin (so any set of apps can share one file), these are all the differences from what the project documents:
- Removed the documented '8096:8096/tcp' and '7359:7359/udp' ports. 7359/udp is only for client auto-discovery on your LAN, so with LAN discovery off apps must be pointed at the address by hand. The LAN discovery switch above the compose file gives the file that publishes it.
- Removed the documented '8096:8096/tcp' port; Caddy reaches jellyfin:8096 on the proxy network. LAN discovery is on: the documented '7359:7359/udp', client auto-discovery on your LAN, is published beside Caddy.
- The documented /path/to/config, /path/to/cache and /path/to/media placeholders become folders next to the compose file; the optional second media folder, fonts folder and 'user:' line are left out.
Notes for Jellyfin behind the proxy
- In Jellyfin: Dashboard > Networking > Known proxies, add 172.30.0.2 (Caddy's fixed address). Jellyfin ignores X-Forwarded-For from anything not listed, so without it every visitor looks like Caddy and remote-access limits do not work.
- Put your media in ./jellyfin/media before the first scan. The setup wizard is open to whoever reaches it first, so finish it straight away.
- Hardware transcoding needs the GPU passed into the container, and this file (None, under Your hardware above) passes none, so Jellyfin transcodes on the processor. Pick Intel / AMD or NVIDIA there for the lines Jellyfin documents.
- Hardware transcoding with Intel / AMD: the file under Your hardware adds group_add '122' (change it to your host's render group id) and the device /dev/dri/renderD128. Then, in Jellyfin: “Enable QSV or VA-API in Jellyfin and uncheck the unsupported codecs.” Jellyfin docs: Hardware acceleration on Intel GPU (Official Docker) checked 2026-10-09
- Hardware transcoding with NVIDIA: the file under Your hardware adds runtime: nvidia and a deploy: reservation for every NVIDIA GPU (count: all). Then, in Jellyfin: “Enable NVENC in Jellyfin and uncheck the unsupported codecs.” Jellyfin docs: Hardware acceleration on NVIDIA GPU (Official Docker) checked 2026-10-09
How to run it
- At your DNS provider, add an A record named
mediaon your domain pointing at the server's public IP address, and forward TCP 80, TCP 443 and UDP 443 to it. - Save the three files in one folder:
docker-compose.ymland.envside by side, the Caddyfile at./caddy/Caddyfile. Replaceexample.orgwith your domain in the Caddyfile, the compose file and the .env, or let the builder do it. - Nothing in
.envneeds filling in;chmod 600 .envall the same. - Start it:
docker compose up -d docker compose logs -f caddy # watch the certificate arrive
- Open the new address (
media.plus your domain) in a browser and follow the notes above.
Back it up
Jellyfin documents two methods: its built-in backup (Dashboard > Backups, Jellyfin 10.11 and newer), which zips the database and chosen metadata while it runs, and a manual copy of its data and config directories with Jellyfin stopped. In the official image both directories are /config.
Below is the BACKUP.md the builder writes next to the compose file above: which of its folders and volumes hold data, what to copy with which container stopped, and a restore check. Pick more apps in the builder and it covers them all in one file.
# BACKUP.md built at dankhost.com for: Jellyfin
What to copy from the docker-compose.yml built with this file, the database dump each project documents (rewritten for the service names in that file), what to copy with which container stopped, and how to check the result. Run every command in the folder that holds docker-compose.yml.
The folders are owned by the containers' users: put sudo in front of tar if it says "Permission denied". Named volumes live under /var/lib/docker/volumes, not next to the compose file, so they are copied through a throwaway ubuntu container (Docker's documented --volumes-from method). tar makes a full copy each run; for large folders an incremental tool (restic, borg, rsync) saves time and space.
Never run "docker compose down -v" on this file: it deletes the named volumes of every app in it.
## Before you start
BACKUP=/mnt/backup/selfhost # change this: a disk or machine other than this server's system disk
mkdir -p "$BACKUP"
## Every time: the three stack files
tar czf "$BACKUP/stack-files.tar.gz" docker-compose.yml .env caddy
.env holds the database passwords and secret keys the apps were set up with; keep this archive as private as the data.
## Caddy (the reverse proxy)
Caddy's docs: 'The data directory must not be treated as a cache.' It holds the TLS certificates and their private keys; without it Caddy has to ask Let's Encrypt for every certificate again.
- caddy-data (named volume, /data in caddy)
BACK UP: TLS certificates, private keys, OCSP staples.
- caddy-config (named volume, /config in caddy)
SKIP: the last active configuration, kept for 'caddy run --resume', which this setup does not use: it starts from the Caddyfile.
- ./caddy (/etc/caddy in caddy)
IN stack-files.tar.gz: the Caddyfile, copied with the stack files above.
Copy (Caddy keeps serving while you copy):
docker run --rm --volumes-from caddy -v "$BACKUP":/backup ubuntu tar cvf /backup/caddy-data.tar /data
Restore check:
tar tf "$BACKUP/caddy-data.tar" > /dev/null && echo OK
docker run --rm --volumes-from caddy -v "$BACKUP":/backup ubuntu bash -c "cd /data && tar xvf /backup/caddy-data.tar --strip 1"
Sources:
- Caddy docs: Conventions, Data directory: https://caddyserver.com/docs/conventions#data-directory (checked 2026-09-24)
- Docker docs: Volumes, Back up, restore, or migrate data volumes: https://docs.docker.com/engine/storage/volumes/#back-up-restore-or-migrate-data-volumes (checked 2026-09-24)
## Jellyfin
Jellyfin documents two methods: its built-in backup (Dashboard > Backups, Jellyfin 10.11 and newer), which zips the database and chosen metadata while it runs, and a manual copy of its data and config directories with Jellyfin stopped. In the official image both directories are /config.
What this compose mounts:
- ./jellyfin/config (/config in jellyfin)
BACK UP: Jellyfin's data and config directories: the database, users, libraries, settings and metadata. Built-in backups are written inside it too.
- ./jellyfin/cache (/cache in jellyfin)
SKIP: Jellyfin's cache; the backup guide copies only the data and config directories.
- ./jellyfin/media (/media in jellyfin)
YOUR FILES: your films, series and music. Jellyfin's backups never include them.
YOUR FILES folders hold your own media, not Jellyfin data. If this server has the only copy, back them up like any other files; the steps below leave them out.
1. Stop jellyfin. Jellyfin's guide: stopping it first is 'extremely important, as otherwise the database will be locked and might not be recoverable when restoring'.
docker compose stop jellyfin
2. Copy while jellyfin is stopped:
tar czf "$BACKUP/jellyfin-config.tar.gz" ./jellyfin/config
3. Start it again:
docker compose start jellyfin
Restore check:
tar tzf "$BACKUP/jellyfin-config.tar.gz" > /dev/null && echo OK
Put a copy back (with jellyfin stopped):
tar xzf "$BACKUP/jellyfin-config.tar.gz" # recreates ./jellyfin/config; move the old folder aside first
Jellyfin's manual restore: stop it, move the current folder aside, copy the backup into its place, start it again. Jellyfin cannot downgrade, and this image is 'latest', so note the version (Dashboard) with each backup and restore onto that version.
Sources:
- Jellyfin docs: Backup and Restore: https://jellyfin.org/docs/general/administration/backup-and-restore/ (checked 2026-09-24)
Update it
What docker compose pull does depends only on how each image line in the compose file above is tagged. For Jellyfin as this page writes it:
| Service | Image | How it is pinned | What docker compose pull does |
|---|---|---|---|
caddy | caddy:2 | Major version only (tag 2) | Takes the newest image tagged 2: minor and patch releases of 2, never 3. A new major version needs a tag edit. |
jellyfin | jellyfin/jellyfin | Not pinned: no tag, which means latest | Moves to the newest image published as latest, which can be a new major version. |
What Jellyfin's docs say about updating
Jellyfin's docs: there is no downgrade. The first start of a new version applies its migrations at once and the old version can no longer use the data, so back up before upgrading, above all between major releases such as 10.10 to 10.11.
- The image here has no tag, so it is latest, which Jellyfin says tracks the newest stable release through major and minor version bumps. To move on purpose instead, Jellyfin publishes series tags: 10 (the latest 10.y.z), 10.11 (the latest 10.11.z) and exact versions such as 10.11.0.
- To go back after a bad upgrade: stop Jellyfin, restore the copies of its config and data, set the image to the version the backup was made with, and start it.
The steps
- Back up first, as in Back it up above. An update can migrate Jellyfin's data on its first start; the copy is the way back.
- Download the new images. Running containers are not touched yet:
docker compose pull
- Recreate the containers whose image changed. Volumes and bind-mounted folders stay as they are:
docker compose up -d
- Watch Jellyfin start (Ctrl+C stops following the log, not the app):
docker compose logs -f jellyfin
- Optional: remove the old images the pull left without a tag. This covers every stack on the machine, not only this one:
docker image prune
Locked out? Reset the admin password
Jellyfin's docs give no command or setting that resets a password.
- What they do name: the Forgot Password feature on the login page, and an admin's Reset Password button on another user's profile in the dashboard, after which that user logs in without a password. Source: Jellyfin docs: Troubleshooting, Unlock locked user account read 2026-10-03; Jellyfin docs: Adding and managing users (Password) read 2026-10-03
- If the admin account is locked after failed logins and Forgot Password does not work, the troubleshooting page gives two SQL statements that unlock it in jellyfin.db. Unlocking does not change the password. Source: Jellyfin docs: Troubleshooting, Unlock locked user account read 2026-10-03
Sources
Every page below was fetched on the date shown.
- Jellyfin README checked 2026-09-23 (what it does, what it replaces)
- Jellyfin docs: Hardware Selection (System Memory) checked 2026-09-23 (hardware requirements)
- Jellyfin docs: Hardware Selection checked 2026-09-23 (hardware requirements)
- Jellyfin docs: Hardware Selection (Shared Items, Storage) checked 2026-09-23 (hardware requirements)
- Jellyfin docs: Container install (Docker Compose) checked 2026-09-23 (compose definition and proxy settings, update notes, LAN discovery)
- Jellyfin docs: Reverse proxy (Known Proxies) checked 2026-09-23 (compose definition and proxy settings)
- Jellyfin docs: Caddy checked 2026-09-23 (compose definition and proxy settings)
- Caddy docs: Keep Caddy running (Docker Compose section) checked 2026-09-23 (Caddy)
- Vaultwarden wiki: Caddy with HTTP challenge (uses the caddy:2 image) checked 2026-09-23 (Caddy)
- Jellyfin docs: Backup and Restore checked 2026-09-24 (backup plan, update notes)
- Caddy docs: Conventions, Data directory checked 2026-09-24 (backup plan)
- Docker docs: Volumes, Back up, restore, or migrate data volumes checked 2026-09-24 (backup plan)
- Docker docs: docker compose pull checked 2026-09-30 (update steps)
- Docker docs: docker compose up (recreates changed containers, keeps volumes) checked 2026-09-30 (update steps)
- Docker docs: docker image prune checked 2026-09-30 (update steps)
- Jellyfin docs: Hardware acceleration on Intel GPU (Official Docker) checked 2026-10-09 (Your hardware)
- Jellyfin docs: Hardware acceleration on AMD GPU (Official Docker) checked 2026-10-09 (Your hardware)
- Jellyfin docs: Hardware acceleration on NVIDIA GPU (Official Docker) checked 2026-10-09 (Your hardware)
- Jellyfin docs: Hardware acceleration on Rockchip VPU (Official Docker) checked 2026-10-09 (Your hardware)
- Jellyfin docs: Networking (port bindings, auto-discovery, firewall) checked 2026-10-09 (LAN discovery)