07. API Connection Refused

Checklist Before Assuming a Code Bug

1. Is the server process running? ps aux | grep ollama or ps aux | grep vllm
2. Is the correct port in use? sudo lsof -i :11434
3. Is the firewall allowing connections? sudo ufw status or sudo iptables -L
4. Is the server binding to the right interface? Some servers bind to 127.0.0.1 by default, accepting only local connections.

Common Causes

Wrong Host Binding

Ollama binds to 127.0.0.1 by default. External access requires:

# Check current binding
sudo lsof -i :11434 | grep LISTEN

# If bound to 127.0.0.1, restart with correct host
OLLAMA_HOST="0.0.0.0" ollama serve

Port Already in Use

# Find what's using the port
sudo lsof -i :8080
# Kill the process
sudo kill $(sudo lsof -t -i :8080)

Docker Network Isolation

# Check if container can reach host
docker exec -it <container_id> curl http://host.docker.internal:8080/health

# Run container with host network
docker run --network host your-ai-service

TLS Certificate Issues

# Test HTTP (non-TLS) connectivity
curl http://localhost:11434/api/tags
# Test with verbose output
curl -v http://localhost:11434/api/tags

Local verification checkpoint

Run the smallest example from this chapter in a local workspace and record the package version, runtime, data path, and observed output. If the result depends on model size, vector count, CPU/GPU backend, or available memory, note that constraint beside the exercise so the lesson remains reproducible.

Local verification checkpoint

Run the smallest example from this chapter in a local workspace and record the package version, runtime, data path, and observed output. If the result depends on model size, vector count, CPU/GPU backend, or available memory, note that constraint beside the exercise so the lesson remains reproducible.