Ollama Errors & Fixes

Quick Diagnostic

# Run this to see exactly what's wrong:
devdiff doctor

---

❌ Error: "model 'llama3.1:8b' not found"

What This Means

You have Ollama installed but the specific AI model is not downloaded. Ollama is the engine — models are separate files (2-4GB each).

Fix

# 1. Verify Ollama is running
ollama list
# Should show your installed models (may be empty)

# 2. Pull the model your config expects
ollama pull llama3.2:3b     # ~2GB — fast, works on most machines
# OR
ollama pull llama3.1:8b     # ~4GB — better quality, needs more RAM

# 3. Verify it's installed
ollama list
# Should now show: llama3.2:3b

# 4. Test it works
ollama run llama3.2:3b "Say hello"
# Should respond with text

# 5. Now try DevDiff
devdiff generate

Which Model Should I Use?

Model	Size	RAM Needed	Speed	Quality	Best For
llama3.2:3b	2GB	4GB	Fast	Good	Most users
llama3.1:8b	4GB	8GB	Medium	Better	Complex analysis
codellama:13b	7GB	16GB	Slow	Best	Security reviews

Start with llama3.2:3b — it's the default and works on most machines.

---

❌ Error: "ECONNREFUSED" or "fetch failed"

What This Means

Ollama is not running. The app is installed but the service isn't active.

Fix

Windows

# Check if Ollama is running
Get-Process ollama -ErrorAction SilentlyContinue

# If not running:
# 1. Open Start Menu
# 2. Search "Ollama"
# 3. Click the Ollama app (white llama icon)
# 4. Wait for icon to appear in system tray
# 5. Try again: devdiff generate

# Or start from command line:
& "C:\Program Files\Ollama\ollama.exe" serve

# Verify it's working:
curl http://localhost:11434/api/tags

macOS

# Check if running
ps aux | grep ollama

# Start Ollama
ollama serve

# Or if installed via brew:
brew services start ollama

Linux

# Check if running
systemctl status ollama

# Start Ollama
sudo systemctl start ollama

# Enable on boot
sudo systemctl enable ollama

---

❌ Error: "Ollama not found"

What This Means

Ollama is not installed on your system.

Fix

Windows

1. Download from ollama.com/download/windows
2. Run the installer (OllamaSetup.exe)
3. After install, open PowerShell:

ollama pull llama3.2:3b

macOS

brew install ollama
ollama serve &
ollama pull llama3.2:3b

Linux

curl -fsSL https://ollama.com/install.sh | sh
ollama pull llama3.2:3b

---

❌ Error: "Out of memory" or crash during generation

What This Means

The model is too large for your available RAM.

Fix

# Use smaller model
ollama pull llama3.2:1b    # Only ~1GB
ollama pull llama3.2:3b    # ~2GB

# Update config to use smaller model
# .devdiff.config.js
export default {
  ai: {
    providers: [
      { name: 'local-small', url: 'ollama://llama3.2:1b', priority: 1 }
    ]
  }
}

---

✅ Verify Everything Works

# Complete health check
devdiff doctor

# Expected output:
# ✅ Node.js v22.4.1
# ✅ Git 2.45.0
# ✅ Ollama installed: v0.3.0
# ✅ Ollama running: http://localhost:11434
# ✅ Models available: llama3.2:3b
# ✅ Model responds: "Hello from Ollama!"
# ✅ DevDiff config valid
# ✅ Git repository detected
# ✅ Staged changes: 3 files
#
# 🎉 All checks passed! Run: devdiff generate