appendix.md

Appendix

Extra information that doesn't fit in the main documentation.

Make Commands

The project includes several make commands to simplify common operations. These commands handle everything from setup to monitoring GPU usage.

Setup Commands

• make install-docker - Installs Docker and all required dependencies
• make setup-gpu - Installs NVIDIA drivers and GPU dependencies
• make setup-watch-tcp - Sets up the logging chain required for TCP connection monitoring

Status Commands

• make test-gpu - Runs a test container to verify GPU availability and driver installation
• make run-status-check - Runs a test container to verify GPU availability and driver installation

Start Commands

• make docker-up-linux-gpu - Starts containers with GPU support on Linux.

  Equivalent to: docker compose -f docker-compose.yml -f docker-compose.gpu.yml up -d

• make docker-up-mac-gpu - Starts containers with GPU support on macOS.

  Equivalent to: docker compose -f docker-compose.yml -f docker-compose.mac.yml up -d

• make docker-up-windows-gpu - Starts containers with GPU support on Windows.

  Equivalent to: docker compose -f docker-compose.yml -f docker-compose.gpu.yml -f docker-compose.windows.yml up -d

• make docker-up-cpu-only - Starts containers without GPU support (CPU only mode).

  Equivalent to: docker compose up -d

Monitoring Commands

• make watch-gpu - Shows real-time GPU usage statistics (updates every 0.5 seconds)
• make watch-tcp-raw - Shows raw TCP connection data from containers in real-time
• make watch-tcp - Shows color-coded TCP connection data from containers in real-time
• make watch-tcp-external-only - Shows color-coded TCP connection data from containers to the outside world in real-time

Utility Commands

• make clean-up - Stops and removes all running containers

FAQ

How do I see if the service is running?

Run docker ps on your host machine. You should see something like this:

How do I check if the model is downloaded?

Run docker logs -f deepseek-ollama to see the progress of the model download.

How do I determine if my GPU is being used?

Method 1

Run make watch-gpu on your host machine. Submit a prompt using the web ui or cli. You should see the processes as below:

If you see something like this, your GPU is being used:

Method 2

Run docker logs deepseek-ollama on your host machine. You should see: