sciagent/be0/TROUBLESHOOTING_CHAT.md

Chat Assistant Troubleshooting Guide

Common Errors and Solutions

Error: 500 Internal Server Error

This usually indicates one of the following issues:

1. Ollama Not Running

Symptoms:

• 500 error on /api/v1/chat
• Error message mentions "connection" or "refused"

Solution:

# Check if Ollama is running in the container
docker exec be0 ps aux | grep ollama

# If not running, restart the container
docker-compose restart be0

# Or start Ollama manually
docker exec be0 ollama serve &

2. Model Not Available

Symptoms:

• Error mentions "model not found"
• Model name mismatch

Solution:

# Check available models
docker exec be0 ollama list

# Pull the required model
docker exec be0 ollama pull gemma3:270M

# Verify model is available
docker exec be0 ollama list | grep gemma3

3. Model Name Mismatch

Issue: Code uses gemma3:27b but entrypoint pulls gemma3:270M

Solution: The code has been updated to use gemma3:270M to match the entrypoint script.

4. Network Connectivity

Symptoms:

• Connection refused errors
• Timeout errors

Solution:

# Check if Ollama is accessible from within the container
docker exec be0 curl http://localhost:11434/api/tags

# Check Ollama service status
docker exec be0 ollama list

Diagnostic Endpoints

Health Check

curl http://localhost:4402/health

This will show:

• Overall service status
• Ollama connection status
• Available models

Test Ollama Directly

# From inside the container
docker exec be0 ollama run gemma3:270M "Hello"

Debugging Steps

1. Check Backend Logs:

   docker-compose logs be0 | tail -50
   

2. Check Chat Assistant Logs:

   tail -f be0/logs/ChatAssistant.log
   

3. Test API Endpoint:

   curl -X POST http://localhost:4402/api/v1/chat \
     -H "Content-Type: application/json" \
     -d '{"message": "Hello"}'
   

4. Verify Ollama Service:

   docker exec be0 ollama list
   docker exec be0 curl http://localhost:11434/api/tags
   

Common Fixes

Fix 1: Restart Ollama Service

docker exec be0 pkill ollama
docker exec be0 ollama serve &
sleep 2
docker exec be0 ollama list

Fix 2: Pull Missing Model

docker exec be0 ollama pull gemma3:270M

Fix 3: Restart Container

docker-compose restart be0

Fix 4: Rebuild Container

docker-compose down
docker-compose build be0
docker-compose up be0

Expected Behavior

When working correctly:

1. Health endpoint shows Ollama as "connected"
2. Available models list includes gemma3:270M
3. Chat endpoint returns 200 with a response
4. Logs show successful message processing

Still Having Issues?

1. Check the full error in logs: docker-compose logs be0
2. Verify Ollama is running: docker exec be0 ps aux | grep ollama
3. Test Ollama directly: docker exec be0 ollama run gemma3:270M "test"
4. Check model availability: docker exec be0 ollama list