Operator Deck

Ollama Command Bridge

Built for tinkerers who break things on purpose. Run diagnostics, simulate recovery steps, and pressure-test your local or remote inference path.

systemd Model Relocation OpenWebUI Link Failure Drills

Live Container Lab

This page now launches a real ephemeral container. Practice commands in isolation, then apply the same workflow on your persistent Ollama host.

Connecting to lab engine...

Checking if real Linux environments are available

Challenges

Validate your live container before installing any runtime tooling.

○ Check the operating system release

○ Verify network connectivity

○ Confirm shell and coreutils are available

○ Check package manager availability

○ Check kernel version

Download, verify, and prepare the Ollama binary for use.

○ Download the Ollama install script

○ Verify the Ollama binary exists

○ Confirm Ollama is in PATH

○ Check Ollama version

○ Check runtime dependencies

Configure Ollama as a managed system service with custom overrides.

○ Inspect the service unit file

○ Set environment variables via override

○ Configure the model storage path

○ Start and stop the service

○ View service logs

Pull, list, inspect, and run inference against local models.

○ Pull a model from the registry

○ List all local models

○ Inspect model storage location

○ Query the API tags endpoint

○ Run a test inference

Bind Ollama to the network and secure remote access.

○ Change the listener binding

○ Configure firewall rules for Ollama

○ Test remote access from another host

○ Set up access control lists

○ Verify remote connectivity with curl

Connect OpenWebUI to a remote Ollama instance and verify model visibility.

○ Understand the base URL format

○ Test the /api/tags endpoint directly

○ Configure the connection in OpenWebUI

○ Troubleshoot connection failures

○ Verify the model list appears in OpenWebUI

Benchmark and optimize Ollama inference performance.

○ Benchmark cold start time

○ Measure inference latency

○ Configure GPU layer allocation

○ Compare model sizes and speed tradeoffs

○ Adjust context window size

Build resilience with crash recovery, health checks, and runbooks.

○ Simulate crash recovery

○ Test mount dependency handling

○ Practice graceful restart

○ Write a health check script

○ Document a recovery runbook

0% Complete

LIVE Connecting...

Lab Engine Status

Checking

Active sessions --

Capacity --

Template --

Last check --

Waiting for lab engine response...

Command Deck

Tap a check to replay output in the terminal pane.

Service health

Topology Pulse

Linux Workstation

Ollama daemon + local models

Healthy

Remote Inference Host

GPU-backed Ollama endpoint

Check ACL

OpenWebUI

Client over HTTP bridge

Connected

Failure Matrix

No models listed

Likely cause: Incorrect OLLAMA_MODELS path or mount not ready

Fast fix: Verify override + RequiresMountsFor, then daemon-reload + restart

OpenWebUI cannot connect

Likely cause: Wrong base URL (often includes /api)

Fast fix: Use host:11434 exactly and validate with curl /api/tags

Service exits on restart

Likely cause: Permission mismatch or inaccessible model directory

Fast fix: Check journalctl -u ollama and directory ownership

Hard Mode Missions

0/5 complete

Relocate models to mounted storage and confirm boot-safe mount dependency. Expose Ollama to LAN intentionally and verify firewall scope. Connect OpenWebUI to remote Ollama and verify model list appears in UI. Run a cold-restart test and confirm no missing-model errors. Capture one benchmark prompt before and after tuning for comparison.

Ollama Command Bridge

Live Container Lab

Real Container Lab

Lab Engine Status

Command Deck

Topology Pulse

Linux Workstation

Remote Inference Host

OpenWebUI

Failure Matrix

No models listed

OpenWebUI cannot connect

Service exits on restart

Hard Mode Missions

System Status

🌐 Gateway

🚀 Orchestrators

🤖 Build Drones

🔨 Active Builds