Skip to content

Add Open WebUI install script - #82

Open
Sportarr wants to merge 2 commits into
eshtek:stagingfrom
Sportarr:add-open-webui
Open

Sportarr wants to merge 2 commits into
eshtek:stagingfrom
Sportarr:add-open-webui

Conversation

@Sportarr

@Sportarr Sportarr commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor

What it does

Open WebUI is a chat interface for AI models. This script wraps the TrueNAS community app open-webui and picks its image with Ollama bundled, so it works on its own with no other app installed.

How it installs

There are no questions. The first person to open it creates the admin account, which is Open WebUI's own first-run step. Models are downloaded from inside the app.

  • image_selector is ollama_image, which runs Open WebUI and Ollama in one container, on CPU or NVIDIA. gpus is $GPU_CONFIG().
  • webui_secret_key and redis_password come from $RANDOM_STRING(32).
  • App data is on ApplicationsPerformance at open-webui/data and is snapshotted before updates. Ollama models go to ApplicationsCapacity at open-webui/ollama, because they are large and can be downloaded again. The script sets no owner, because the container runs as root.
  • The port is 31028, the TrueNAS default.
  • Limits are $CPUS(75%, 2) and $MEMORY(50%, 4096). The memory limit covers Ollama too, so loaded models have to fit inside it.
  • The requirements list 8GBRAM, since loaded models need memory. I left out GPU because no other script uses it, and your schema page describes it as required hardware.

Testing

  • bun _lib/validate.ts passes with 0 errors and 0 warnings.
  • I installed it on my HexOS server (TrueNAS 25.10, HexOS 2026.10.16), with 14 cores and 15 GB of memory. The macros gave 10 CPUs, a 7663 MB memory limit and use_all_gpus for the Intel GPU. The bundled Ollama started and reported the limit as its memory. /health returned 200, and /api/config reported onboarding: true, so the first visit goes to admin account creation. Idle, it used about 900 MB.

@Sportarr

Copy link
Copy Markdown
Contributor Author

I just noticed your feature/ai-apps branch already has an Open WebUI script, along with Ollama, LocalAI and AnythingLLM. Your branch and this PR both add open-webui.json, so only one of them can land. Yours points Open WebUI at the separate Ollama app. This one uses TrueNAS's image with Ollama built in and passes the GPU through. If your AI apps work covers Open WebUI, I'm happy to close this one. If the built-in Ollama setup or anything else here helps, feel free to take it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant