Open WebUI with Docker Compose
Open WebUI is a self-hosted AI platform that's extensible, feature-rich, user-friendly, and built to run entirely offline. With support for Ollama and OpenAI-compatible APIs, it works with both local and cloud-based models. In the project's words: Open WebUI docs: front page checked 2026-09-25
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 Open WebUI to the builder to use your own domain, get its passwords generated in your browser, and combine it with other apps in one file.
What Open WebUI needs
Only figures Open WebUI'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 | The project publishes no figure | |
| CPU | The project publishes no figure | |
| Disk | The project publishes no figure | |
Which hardware Open WebUI 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 forollama,open-webui. No image for this hardware; the builder leaves the app out.
| Service | Image | Platforms in the manifest | Digest | Registry | Checked |
|---|---|---|---|---|---|
ollama | ollama/ollama:${OLLAMA_DOCKER_TAG-latest} | amd64, arm64 | sha256:8262851b2846b87c649eddf3e76beb270c52f4d1bc94559f47efde16b0841551 | registry-1.docker.io | 2026-09-25 |
open-webui | ghcr.io/open-webui/open-webui:${WEBUI_DOCKER_TAG-main} | amd64, arm64 | sha256:8b432fe0a65b91116afc7961365c6cca5379cc923171386a96691f3471f3cae9 | ghcr.io | 2026-09-25 |
Your hardware
Open WebUI's file runs Ollama, the program that runs the models, with no GPU passed into its container, so every reply is computed on the processor. Pick what the server has: the file below then gives the ollama service the lines Open WebUI's repository and Ollama's Docker docs give for that card, each quoted with its source, and names what the host needs first. The open-webui service itself does not change.
docker-compose.yml
CPU only: models run on the processor
This is exactly what the builder writes for Open WebUI alone on example.org: the official definition with its published ports removed and Caddy added in front. Every ${VARIABLE} is defined in the .env below.
# docker-compose.yml built at dankhost.com for: Open WebUI # Each app block is its official compose definition. Only the names were prefixed with # the app slug (services, containers, volumes, bind folders, .env variables) so any set of # apps can share one file; other differences are listed under "Changed:" in each block. # Only Caddy publishes ports. Secrets live in .env next to this file. services: # ---- Caddy: the one reverse proxy, HTTPS certificates from Let's Encrypt ---- # Source: https://caddyserver.com/docs/running (checked 2026-09-23) # Source: https://github.com/dani-garcia/vaultwarden/wiki/Using-Docker-Compose (checked 2026-09-23) # Caddyfile goes in ./caddy/Caddyfile. Fixed address 172.30.0.2 so apps can trust it exactly. caddy: image: "caddy:2" container_name: "caddy" restart: "unless-stopped" ports: - "80:80" - "443:443" - "443:443/udp" volumes: - "./caddy:/etc/caddy" - "caddy-data:/data" - "caddy-config:/config" networks: proxy: ipv4_address: "172.30.0.2" # ---- Open WebUI (chat.example.org) ---- # Source: https://github.com/open-webui/open-webui/blob/main/docker-compose.yaml (checked 2026-09-25) # Source: https://docs.openwebui.com/getting-started/quick-start/ (checked 2026-09-25) # Changed: Removed the '${OPEN_WEBUI_PORT-3000}:8080' port; Caddy reaches open-webui:8080 on the proxy network. Ollama publishes no port at all: only Open WebUI talks to it, as http://ollama:11434 on the app's own network. # Removed the 'build:' section (it builds the image from a source checkout) so the published image is pulled, and the documented container names 'ollama' and 'open-webui' (the builder names every container after its service). # WEBUI_SECRET_KEY is a generated 64-character key instead of the empty value in the file: the Quick Start says to set it once and keep it, because without a fixed key every recreated container logs everyone out. # Removed extra_hosts host.docker.internal:host-gateway: it lets Open WebUI reach an Ollama running on the host, and here Ollama runs in its own container. # Ollama's network keeps internet access (unlike the private database networks of other apps) so that it can download models. open-webui-ollama: image: "ollama/ollama:${OPEN_WEBUI_OLLAMA_DOCKER_TAG-latest}" container_name: "open-webui-ollama" volumes: - "open-webui-ollama:/root/.ollama" pull_policy: "always" tty: true restart: "unless-stopped" networks: open-webui-backend: aliases: - "ollama" open-webui: image: "ghcr.io/open-webui/open-webui:${OPEN_WEBUI_WEBUI_DOCKER_TAG-main}" container_name: "open-webui" volumes: - "open-webui:/app/backend/data" depends_on: - "open-webui-ollama" environment: OLLAMA_BASE_URL: "http://ollama:11434" WEBUI_SECRET_KEY: "${OPEN_WEBUI_WEBUI_SECRET_KEY}" restart: "unless-stopped" networks: proxy: {} open-webui-backend: {} 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" open-webui-backend: {} volumes: caddy-data: caddy-config: open-webui-ollama: open-webui:
With NVIDIA
An NVIDIA card, with the NVIDIA driver and the NVIDIA Container Toolkit on the host. For an NVIDIA card: the deploy: GPU reservation of Open WebUI's docker-compose.gpu.yaml (driver: nvidia, count: 1, capabilities: [gpu]) added to the ollama service. That file reads the driver and the count from OLLAMA_GPU_DRIVER and OLLAMA_GPU_COUNT, whose defaults are nvidia and 1; this file writes the defaults. Everything else is the CPU only file, as the builder writes it for Open WebUI alone on example.org.
On the host first: “Install the NVIDIA Container Toolkit.” Ollama v0.40.1 docs: Docker checked 2026-10-09
# docker-compose.yml built at dankhost.com for: Open WebUI # 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" # ---- Open WebUI (chat.example.org) ---- # Source: https://github.com/open-webui/open-webui/blob/main/docker-compose.yaml (checked 2026-09-25) # Source: https://docs.openwebui.com/getting-started/quick-start/ (checked 2026-09-25) # Source: https://github.com/open-webui/open-webui/blob/v0.11.4/docker-compose.yaml (checked 2026-10-09) # Source: https://github.com/open-webui/open-webui/blob/v0.11.4/docker-compose.gpu.yaml (checked 2026-10-09) # Source: https://github.com/ollama/ollama/blob/v0.40.1/docs/docker.mdx (checked 2026-10-09) # Source: https://github.com/ollama/ollama/blob/v0.40.1/docs/gpu.mdx (checked 2026-10-09) # Changed: Removed the '${OPEN_WEBUI_PORT-3000}:8080' port; Caddy reaches open-webui:8080 on the proxy network. Ollama publishes no port at all: only Open WebUI talks to it, as http://ollama:11434 on the app's own network. # Removed the 'build:' section (it builds the image from a source checkout) so the published image is pulled, and the documented container names 'ollama' and 'open-webui' (the builder names every container after its service). # WEBUI_SECRET_KEY is a generated 64-character key instead of the empty value in the file: the Quick Start says to set it once and keep it, because without a fixed key every recreated container logs everyone out. # Removed extra_hosts host.docker.internal:host-gateway: it lets Open WebUI reach an Ollama running on the host, and here Ollama runs in its own container. # Ollama's network keeps internet access (unlike the private database networks of other apps) so that it can download models. # For an NVIDIA card: the deploy: GPU reservation of Open WebUI's docker-compose.gpu.yaml (driver: nvidia, count: 1, capabilities: [gpu]) added to the ollama service. That file reads the driver and the count from OLLAMA_GPU_DRIVER and OLLAMA_GPU_COUNT, whose defaults are nvidia and 1; this file writes the defaults. open-webui-ollama: image: "ollama/ollama:${OPEN_WEBUI_OLLAMA_DOCKER_TAG-latest}" container_name: "open-webui-ollama" volumes: - "open-webui-ollama:/root/.ollama" pull_policy: "always" tty: true restart: "unless-stopped" deploy: resources: reservations: devices: - driver: "nvidia" count: 1 capabilities: - "gpu" networks: open-webui-backend: aliases: - "ollama" open-webui: image: "ghcr.io/open-webui/open-webui:${OPEN_WEBUI_WEBUI_DOCKER_TAG-main}" container_name: "open-webui" volumes: - "open-webui:/app/backend/data" depends_on: - "open-webui-ollama" environment: OLLAMA_BASE_URL: "http://ollama:11434" WEBUI_SECRET_KEY: "${OPEN_WEBUI_WEBUI_SECRET_KEY}" restart: "unless-stopped" networks: proxy: {} open-webui-backend: {} networks: # Pinned so the trusted-proxy address given to each app (172.30.0.2) is exact. # If "docker compose up" reports "Pool overlaps", change 172.30.0 here, in Caddy's # ipv4_address and in every trusted-proxy value to a free 172.x.0 range. proxy: ipam: config: - subnet: "172.30.0.0/24" ip_range: "172.30.0.128/25" gateway: "172.30.0.1" open-webui-backend: {} volumes: caddy-data: caddy-config: open-webui-ollama: open-webui:
ollama/ollama:${OLLAMA_DOCKER_TAG-latest}: “ image: ollama/ollama:${OLLAMA_DOCKER_TAG-latest}” Open WebUI v0.11.4: docker-compose.yaml checked 2026-10-09nvidia,1,gpu: Open WebUI v0.11.4: docker-compose.gpu.yaml checked 2026-10-09- driver: ${OLLAMA_GPU_DRIVER-nvidia} count: ${OLLAMA_GPU_COUNT-1} capabilities: - gpu- Then, per the same page, point Docker at it and restart Docker: Ollama v0.40.1 docs: Docker checked 2026-10-09
sudo nvidia-ctk runtime configure --runtime=docker sudo systemctl restart docker
- Which cards and drivers, per Ollama's hardware page: “Ollama supports Nvidia GPUs with compute capability 5.0+ and driver version 550 and newer.” Ollama v0.40.1 docs: Hardware support checked 2026-10-09
- Ollama's own docker run gives the container every GPU (--gpus=all); Open WebUI's file, and so this one, reserves one: “docker run -d --gpus=all -v ollama:/root/.ollama -p 11434:11434 --name ollama ollama/ollama” Ollama v0.40.1 docs: Docker checked 2026-10-09
With AMD
An AMD card that ROCm supports, with AMD's ROCm driver on the host. For an AMD card: the ollama/ollama:rocm image and the devices /dev/kfd and /dev/dri on the ollama service, as in Ollama's docker run for AMD and Open WebUI's docker-compose.amdgpu.yaml. Open WebUI's file writes the tag as ${OLLAMA_DOCKER_TAG-rocm}, but this file's .env sets OPEN_WEBUI_OLLAMA_DOCKER_TAG=latest for the CPU only file, so the rocm tag is written out. Open WebUI's file also sets HSA_OVERRIDE_GFX_VERSION, which this file leaves out (see the notes below). Everything else is the CPU only file, as the builder writes it for Open WebUI alone on example.org.
On the host first: “Ollama requires the AMD ROCm v7 driver on Linux.” Ollama v0.40.1 docs: Hardware support checked 2026-10-09
# docker-compose.yml built at dankhost.com for: Open WebUI # 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" # ---- Open WebUI (chat.example.org) ---- # Source: https://github.com/open-webui/open-webui/blob/main/docker-compose.yaml (checked 2026-09-25) # Source: https://docs.openwebui.com/getting-started/quick-start/ (checked 2026-09-25) # Source: https://github.com/ollama/ollama/blob/v0.40.1/docs/docker.mdx (checked 2026-10-09) # Source: https://github.com/open-webui/open-webui/blob/v0.11.4/docker-compose.amdgpu.yaml (checked 2026-10-09) # Source: https://github.com/ollama/ollama/blob/v0.40.1/docs/gpu.mdx (checked 2026-10-09) # Changed: Removed the '${OPEN_WEBUI_PORT-3000}:8080' port; Caddy reaches open-webui:8080 on the proxy network. Ollama publishes no port at all: only Open WebUI talks to it, as http://ollama:11434 on the app's own network. # Removed the 'build:' section (it builds the image from a source checkout) so the published image is pulled, and the documented container names 'ollama' and 'open-webui' (the builder names every container after its service). # WEBUI_SECRET_KEY is a generated 64-character key instead of the empty value in the file: the Quick Start says to set it once and keep it, because without a fixed key every recreated container logs everyone out. # Removed extra_hosts host.docker.internal:host-gateway: it lets Open WebUI reach an Ollama running on the host, and here Ollama runs in its own container. # Ollama's network keeps internet access (unlike the private database networks of other apps) so that it can download models. # For an AMD card: the ollama/ollama:rocm image and the devices /dev/kfd and /dev/dri on the ollama service, as in Ollama's docker run for AMD and Open WebUI's docker-compose.amdgpu.yaml. Open WebUI's file writes the tag as ${OLLAMA_DOCKER_TAG-rocm}, but this file's .env sets OPEN_WEBUI_OLLAMA_DOCKER_TAG=latest for the CPU only file, so the rocm tag is written out. Open WebUI's file also sets HSA_OVERRIDE_GFX_VERSION, which this file leaves out (see the notes below). open-webui-ollama: image: "ollama/ollama:rocm" container_name: "open-webui-ollama" devices: - "/dev/kfd:/dev/kfd" - "/dev/dri:/dev/dri" volumes: - "open-webui-ollama:/root/.ollama" pull_policy: "always" tty: true restart: "unless-stopped" networks: open-webui-backend: aliases: - "ollama" open-webui: image: "ghcr.io/open-webui/open-webui:${OPEN_WEBUI_WEBUI_DOCKER_TAG-main}" container_name: "open-webui" volumes: - "open-webui:/app/backend/data" depends_on: - "open-webui-ollama" environment: OLLAMA_BASE_URL: "http://ollama:11434" WEBUI_SECRET_KEY: "${OPEN_WEBUI_WEBUI_SECRET_KEY}" restart: "unless-stopped" networks: proxy: {} open-webui-backend: {} networks: # Pinned so the trusted-proxy address given to each app (172.30.0.2) is exact. # If "docker compose up" reports "Pool overlaps", change 172.30.0 here, in Caddy's # ipv4_address and in every trusted-proxy value to a free 172.x.0 range. proxy: ipam: config: - subnet: "172.30.0.0/24" ip_range: "172.30.0.128/25" gateway: "172.30.0.1" open-webui-backend: {} volumes: caddy-data: caddy-config: open-webui-ollama: open-webui:
ollama/ollama:rocm: “docker run -d --device /dev/kfd --device /dev/dri -v ollama:/root/.ollama -p 11434:11434 --name ollama ollama/ollama:rocm” Ollama v0.40.1 docs: Docker checked 2026-10-09/dev/kfd:/dev/kfd,/dev/dri:/dev/dri: Open WebUI v0.11.4: docker-compose.amdgpu.yaml checked 2026-10-09devices: - /dev/kfd:/dev/kfd - /dev/dri:/dev/dri
- Ollama's Docker page on AMD: “To run Ollama using Docker with AMD GPUs, use the `rocm` tag and the following command:” Ollama v0.40.1 docs: Docker checked 2026-10-09
- Open WebUI's AMD file names the same image through a variable: “ image: ollama/ollama:${OLLAMA_DOCKER_TAG-rocm}” Open WebUI v0.11.4: docker-compose.amdgpu.yaml checked 2026-10-09
- Open WebUI's AMD file also sets this, which has ROCm treat any card as GFX version 11.0.0. Ollama's hardware page describes the variable as a way to run a card ROCm does not support, so this file leaves it out; add it under environment only if your card is not on that page's list: “ - 'HSA_OVERRIDE_GFX_VERSION=${HSA_OVERRIDE_GFX_VERSION-11.0.0}'” Open WebUI v0.11.4: docker-compose.amdgpu.yaml checked 2026-10-09
- Ollama's hardware page on that variable: “Ollama leverages the AMD ROCm library, which does not support all AMD GPUs. In some cases you can force the system to try to use a similar LLVM target that is close.” Ollama v0.40.1 docs: Hardware support checked 2026-10-09
- For AMD cards outside ROCm, Ollama's hardware page notes: “Additional AMD GPU support is provided by the Vulkan Library” Ollama v0.40.1 docs: Hardware support checked 2026-10-09
- and its Docker page says of the plain ollama/ollama image (the CPU only file's) with the same two devices: “Vulkan is bundled into the `ollama/ollama` image and is enabled by default when the container can access the GPU devices.” Ollama v0.40.1 docs: Docker checked 2026-10-09
- If SELinux keeps the container from the GPU: “On the host system you can run `sudo setsebool container_use_devices=1` to allow containers to use devices.” Ollama v0.40.1 docs: Hardware support checked 2026-10-09
Caddyfile save as caddy/Caddyfile
Caddy gets and renews the HTTPS certificate for chat.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.
# Open WebUI
chat.example.org {
reverse_proxy open-webui:8080
}
.env
The secret is left blank here on purpose: a password printed on a public page is the same for everyone who copies it. The builder fills it with random letters and digits made in your browser.
# .env for Open WebUI, from dankhost.com/app/open-webui/. Fill the blank secret before starting. # Keep this file private (chmod 600). # Open WebUI # 64 random letters and digits: the builder fills this in your browser OPEN_WEBUI_WEBUI_SECRET_KEY= OPEN_WEBUI_OLLAMA_DOCKER_TAG=latest OPEN_WEBUI_WEBUI_DOCKER_TAG=main
Changes from the official file
Besides prefixing every service, container, volume, folder and variable with open-webui (so any set of apps can share one file), these are all the differences from what the project documents:
- Removed the '${OPEN_WEBUI_PORT-3000}:8080' port; Caddy reaches open-webui:8080 on the proxy network. Ollama publishes no port at all: only Open WebUI talks to it, as http://ollama:11434 on the app's own network.
- Removed the 'build:' section (it builds the image from a source checkout) so the published image is pulled, and the documented container names 'ollama' and 'open-webui' (the builder names every container after its service).
- WEBUI_SECRET_KEY is a generated 64-character key instead of the empty value in the file: the Quick Start says to set it once and keep it, because without a fixed key every recreated container logs everyone out.
- Removed extra_hosts host.docker.internal:host-gateway: it lets Open WebUI reach an Ollama running on the host, and here Ollama runs in its own container.
- Ollama's network keeps internet access (unlike the private database networks of other apps) so that it can download models.
Notes for Open WebUI behind the proxy
- Open https://chat.example.org and sign up at once: the first account created becomes the administrator.
- Ollama starts with no model. Pull one before your first chat, for example: docker compose exec open-webui-ollama ollama pull llama3.2. The model you pull, not Open WebUI, decides how much RAM and disk this needs; neither project's install page gives a figure.
- Ollama runs the models on the processor in this file (CPU only, under Your hardware above): no GPU is passed into its container. Pick NVIDIA or AMD there for the lines Open WebUI's repository and Ollama's Docker docs give.
- NVIDIA: the file under Your hardware adds a deploy: reservation of one NVIDIA GPU (driver: nvidia, count: 1, capabilities: [gpu]) on the ollama service. On the host first: “Install the NVIDIA Container Toolkit.” Ollama v0.40.1 docs: Docker checked 2026-10-09
- AMD: the file under Your hardware adds the ollama/ollama:rocm image and the devices /dev/kfd and /dev/dri on the ollama service. On the host first: “Ollama requires the AMD ROCm v7 driver on Linux.” Ollama v0.40.1 docs: Hardware support checked 2026-10-09
How to run it
- At your DNS provider, add an A record named
chaton 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. - Fill the blank
OPEN_WEBUI_WEBUI_SECRET_KEYin.env. The builder generates it; on the server,tr -dc A-Za-z0-9 </dev/urandom | head -c 32prints 32 random letters and digits. Thenchmod 600 .env. - Start it:
docker compose up -d docker compose logs -f caddy # watch the certificate arrive
- Open the new address (
chat.plus your domain) in a browser and follow the notes above.
Back it up
Open WebUI's backup page lists what /app/backend/data holds: webui.db (the SQLite database with users, chats and settings), uploads/ (uploaded files), vector_db/ (the ChromaDB vector database) and cache/. Its scripts stop the containers before copying. Ollama's models can be downloaded again.
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: Open WebUI
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 "$BACKUP":/backup ubuntu bash -c "cd /data && tar xvf /backup/caddy-data.tar --strip 1"
Sources:
- Caddy docs: Conventions, Data directory: https://caddyserver.com/docs/conventions#data-directory (checked 2026-09-24)
- Docker docs: Volumes, Back up, restore, or migrate data volumes: https://docs.docker.com/engine/storage/volumes/#back-up-restore-or-migrate-data-volumes (checked 2026-09-24)
## Open WebUI
Open WebUI's backup page lists what /app/backend/data holds: webui.db (the SQLite database with users, chats and settings), uploads/ (uploaded files), vector_db/ (the ChromaDB vector database) and cache/. Its scripts stop the containers before copying. Ollama's models can be downloaded again.
What this compose mounts:
- open-webui-ollama (named volume, /root/.ollama in open-webui-ollama)
OPTIONAL: the downloaded models, often several GB each. They can be pulled again with ollama pull; copy them only to save the download.
- open-webui (named volume, /app/backend/data in open-webui)
BACK UP: webui.db (users, chats, settings), uploads/, vector_db/ and cache/.
1. Stop open-webui. Open WebUI's backup page stops the containers before copying so the SQLite database and the vector store are not written mid-copy.
docker compose stop open-webui
2. Copy while open-webui is stopped:
docker run --rm --volumes-from open-webui -v "$BACKUP":/backup ubuntu tar cvf /backup/open-webui.tar /app/backend/data
3. Start it again:
docker compose start open-webui
Restore check:
tar tf "$BACKUP/open-webui.tar" > /dev/null && echo OK
Put a copy back (with open-webui stopped):
docker run --rm --volumes-from open-webui -v "$BACKUP":/backup ubuntu bash -c "cd /app/backend/data && tar xvf /backup/open-webui.tar --strip 3"
Stop it, put the open-webui volume back, start it, then pull the models again if you did not copy the ollama volume.
Sources:
- Open WebUI docs: Backups (what /app/backend/data holds; stop before copying): https://docs.openwebui.com/tutorials/maintenance/backups (checked 2026-09-25)
- 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 2026-09-25)
Update it
What docker compose pull does depends only on how each image line in the compose file above is tagged. For Open WebUI 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. |
open-webui-ollama | ollama/ollama:${OPEN_WEBUI_OLLAMA_DOCKER_TAG-latest} | Not pinned: moving tag latest (.env sets OPEN_WEBUI_OLLAMA_DOCKER_TAG=latest) | Moves to the newest image published as latest, which can be a new major version. |
open-webui | ghcr.io/open-webui/open-webui:${OPEN_WEBUI_WEBUI_DOCKER_TAG-main} | Not pinned: moving tag main (.env sets OPEN_WEBUI_WEBUI_DOCKER_TAG=main) | Moves to the newest image published as main, which can be a new major version. |
What Open WebUI's docs say about updating
Open WebUI's Quick Start: to update a Compose install, docker compose pull and then docker compose up -d.
- In this file both tags come from .env: OPEN_WEBUI_WEBUI_DOCKER_TAG=main and OPEN_WEBUI_OLLAMA_DOCKER_TAG=latest, both moving tags. To stay on one version, set a version tag in those lines.
The steps
- Back up first, as in Back it up above. An update can migrate Open WebUI'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 Open WebUI start (Ctrl+C stops following the log, not the app):
docker compose logs -f open-webui
- 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
Writes a new bcrypt password hash for one account straight into Open WebUI's SQLite database, webui.db.
- Start a throwaway Alpine container with Open WebUI's data mounted. The docs mount the volume at /data; --volumes-from mounts it where the app has it, /app/backend/data. Source: Open WebUI docs: Reset Admin Password (Alternate Docker Method) read 2026-10-03; Docker docs: Volumes, Back up, restore, or migrate data volumes (--volumes-from) read 2026-10-03
docker run -it --rm --volumes-from open-webui alpine
- Inside it, install the tools: Source: Open WebUI docs: Reset Admin Password (Alternate Docker Method) read 2026-10-03
apk add apache2-utils sqlite
- Make the bcrypt hash of your new password: Source: Open WebUI docs: Reset Admin Password (Alternate Docker Method) read 2026-10-03
htpasswd -bnBC 10 "" '<new password>' | tr -d ':'
- Open the database and write the hash for your admin's email. Emails are stored lower-case, so a mixed-case address matches no row. Ctrl+D leaves sqlite. Source: Open WebUI docs: Reset Admin Password (Alternate Docker Method) read 2026-10-03
sqlite3 /app/backend/data/webui.db UPDATE auth SET password='<hash from step 3>' WHERE email='<admin email>';
Sources
Every page below was fetched on the date shown.
- Open WebUI docs: front page checked 2026-09-25 (what it does)
- Open WebUI repository: docker-compose.yaml (open-webui with an ollama service) checked 2026-09-25 (compose definition and proxy settings)
- Open WebUI docs: Quick Start (Docker Compose, WEBUI_SECRET_KEY, data volume) checked 2026-09-25 (compose definition and proxy settings, update notes)
- 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)
- Open WebUI docs: Backups (what /app/backend/data holds; stop before copying) checked 2026-09-25 (backup plan)
- Docker docs: Volumes, Back up, restore, or migrate data volumes checked 2026-09-25 (backup plan)
- Caddy docs: Conventions, Data directory 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)
- Open WebUI v0.11.4: docker-compose.yaml checked 2026-10-09 (Your hardware)
- Open WebUI v0.11.4: docker-compose.gpu.yaml checked 2026-10-09 (Your hardware)
- Ollama v0.40.1 docs: Docker checked 2026-10-09 (Your hardware)
- Ollama v0.40.1 docs: Hardware support checked 2026-10-09 (Your hardware)
- Open WebUI v0.11.4: docker-compose.amdgpu.yaml checked 2026-10-09 (Your hardware)