Unload idle Ollama models and log GPU usage with HTTP and file writes

by Kevin · 0 views · created 2026-10-04 · DevOps

What this workflow does

This workflow has 10 nodes and runs manually, when you click “Execute workflow” (plus 1 other trigger). It uses only n8n's built-in nodes.

How it runs

Trigger: Manually, when you click “Execute workflow” (plus 1 other trigger).

1. Settings (Edit Fields (Set)): sets 6 fields: ollamaUrl, dryRun, idleMinutes…
2. List loaded models (HTTP Request): calls an HTTP API (GET)
3. Decide idle models (Code): runs custom JavaScript code
4. Changed since last run? (If): checks a condition and branches into true / false
5. Live unload needed? (If): checks a condition and branches into true / false
6. Pack log line (Code): runs custom JavaScript code
7. Unload model in Ollama (HTTP Request): calls an HTTP API (POST)
8. Append guard log (Read/Write Files from Disk): writes a file to disk

Before you upgrade to n8n 3.0

Nodes used (7)

• Manual Triggern8n-nodes-base.manualTrigger
• Schedule Triggern8n-nodes-base.scheduleTrigger
• Edit Fields (Set)n8n-nodes-base.set
• HTTP Requestn8n-nodes-base.httpRequest
• Coden8n-nodes-base.code
• Ifn8n-nodes-base.if
• Read/Write Files from Diskn8n-nodes-base.readWriteFile

From the author

Quick Overview This workflow runs manually or every 10 minutes to check which Ollama models are loaded, log state changes to a file, and optionally unload models that have been idle too long to free GPU memory. How it works 1. Runs on a manual trigger or on a…

View the original on n8n.io → Check your own copy