FlynDice

Do you have a Homebrew Path (mount for persistence) set under additional settings? And then you have to actually install Homebrew (linuxbrew)with : /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

that works here also. Had to add it to "allowedOrigins" also... "allowedOrigins": [ "http://<UNRAID-IP>:18789", "http://127.0.0.1:18789" ],

I never had to approve a browser device... telegram had to be approved but no the browser. I've been using this line under advanced in the docker template: WebUI: http://[IP]:[PORT:18789]/#token=MyReDaCtEdGaTeWaYtOkEn link works perfectly. I notice in a lot of screenshots i see [?token=] vs [#token=] I posted the gateway config I have been using a few days ago.

ghcr.io/openclaw/openclaw:latest works fine for me now. it is co-tagged with 2026.5.2-slim right now. see: OpenClaw tags

There was a bug that was just corrected that had to do with preloading some node packages. If you try loading https://github.com/ghcr.io/openclaw/openclaw:2026.5.2-slim vs :latestin your template you may make out ok...

I solved the gateway loop on startup by removing the plugin-runtime-deps/ dir in .openclaw. Latest build starts up fine for me now & it recreated the dir. here is my gateway config if it helps: "gateway": { "mode": "local", "bind": "lan", "controlUi": { "allowInsecureAuth": true, "allowedOrigins": [ "http://<Unraid IP>:18789" ], "dangerouslyDisableDeviceAuth": true }, "auth": { "mode": "token", "token": "REDACTED", "rateLimit": { "maxAttempts": 10, "windowMs": 60000, "lockoutMs": 300000 } }, "port": 18789, "tailscale": { "mode": "off", "resetOnExit": false } },

Versionv2026.4.25 appears to have the gateway fixed now (:latest) I've added this to my post-install to load QMD also as it appears to not be installed in the container: sh -c "mkdir -p /root/.openclaw /home/linuxbrew; [ -s /root/.openclaw/openclaw.json ] || echo '{\"gateway\":{\"mode\":\"local\",\"bind\":\"lan\",\"controlUi\":{\"allowInsecureAuth\":true},\"auth\":{\"mode\":\"token\"}}}' > /root/.openclaw/openclaw.json; if ! command -v qmd >/dev/null 2>&1; then echo 'Installing QMD...'; npm install -g @tobilu/qmd; echo 'QMD installed.'; fi; exec node dist/index.js gateway --bind lan"

I've got openclaw with local ollama running on my unraid box and my laptop. I tried this because I've got no GPU on the unraid box but my laptop has a ryzen with the Radeon 780M gpu . The performance is comparable between the 2 and now I have 2 parallel lanes... I added them by running "openclaw onboard" several times through the terminal. Good Luck! Here are the model/ollama bits of my config: "apiKey": "OLLAMA_API_KEY", <----- literal, not placeholder, just needs to be something "auth": { "profiles": { "ollama:default": { "provider": "ollama", "mode": "api_key" }, "ollama-laptop:default": { "provider": "ollama-laptop", "mode": "api_key" } } }, "models": { "mode": "merge", "providers": { "ollama": { "baseUrl": "http://<CONTAINER IP>:11434", "api": "ollama", "apiKey": "OLLAMA_API_KEY", "models": [ { "id": "qwen2.5-coder:14b", "name": "qwen2.5-coder:14b", "reasoning": false, "input": [ "text" ], "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, "contextWindow": 32768, "maxTokens": 8192, "compat": { "supportsTools": false } }, { "id": "llama3.1:8b", "name": "llama3.1:8b", "reasoning": false, "input": [ "text" ], "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, "contextWindow": 131072, "maxTokens": 8192 }, { "id": "minicpm-v:8b", "name": "minicpm-v:8b", "reasoning": false, "input": [ "text", "image" ], "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, "contextWindow": 32768, "maxTokens": 8192 } ] }, "ollama-laptop": { "baseUrl": "http://192.168.X.XXX:11434", "api": "ollama", "apiKey": "OLLAMA_API_KEY", "models": [ { "id": "llama3.1:8b", "name": "llama3.1:8b (Laptop)", "reasoning": false, "input": [ "text" ], "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, "contextWindow": 131072, "maxTokens": 8192, "compat": { "requiresStringContent": true } }, { "id": "qwen2.5-coder:7b", "name": "qwen2.5-coder:7b (Laptop)", "reasoning": false, "input": [ "text" ], "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, "contextWindow": 32768, "maxTokens": 8192, "compat": { "requiresStringContent": true, "supportsTools": false } } ] } } }

I've had the openclaw container successfully running for awhile now & had to revert back to version ghcr.io/openclaw/openclaw:2026.4.23-amd64 after the last container upgrade. Didn't bother to troubleshoot yet as I figure this is above my pay grade... Looking around some more... Sorry for the spam if the answer is on the previous six pages but, is there a reason the template is using user=root as opposed to the unprivileged user, "node"? From the openclaw dockerfile(268-271): # Security hardening: Run as non-root user # The node:24-bookworm image includes a 'node' user (uid 1000) # This reduces the attack surface by preventing container escape via root privileges USER node Was that just to make it work? The mapped volume for the workspace looks like it may have been set up for "node" originally but now does nothing as the template uses Extra Parameters: --user root --hostname OpenClaw in the Advanced View page... Things that make you go hmmmm....

Works for me too, back to normal now! 👍

Try clearing your browser cache in chrome. That fixed the noVNC problem for me.

Had similar problem & clearing browser cache resolved this for me.

Unfortunately no difference. Update: I restored to 6.7.2 and ended up with the same problem. After clearing the browser cache noVNC access to the vms was normal in both 6.72 and 6.8 😁

After upgrading to 6.8 I'm now getting this noVNC error while trying to access both vms I run. noVNC encountered an error: Uncaught SyntaxError: The requested module './util/browser.js' does not provide an export named 'dragThreshold' http://192.168.1.4/plugins/dynamix.vm.manager/vnc.html?autoconnect=true&host=192.168.1.4&port=&path=/wsproxy/5700/:0:0 SyntaxError: The requested module './util/browser.js' does not provide an export named 'dragThreshold' I can access the vms just fine with NoMachine and Remmina but not the VNC Remote menu option in Unraid.