Planetensimulation

A semester project to learn about several technologies:

• (Py)Qt5
• Parallelization in Python
• Distribution using Python
• (Redis as a fast in-memory DB)

Setup & Dependencies

Debian (tested with Python 3.6.2)

apt-get's Python3 version is too old

# compile & install python 3.6.2 | is this necessary?
cd /tmp
wget https://www.python.org/ftp/python/3.6.3/Python-3.6.3.tar.xz
tar -xvf Python-3.6.3.tar.xz
cd Python-3.6.3/
./configure
make -j4
sudo make install

sudo apt-get install python3-dev python3-pip
sudo pip3 install numpy numba cython

Archlinux

# worker
pacman -Sy gcc python python-numpy cython python-pyqt5

Windows (tested Windows 7?)

1. Download & install Python 3.6.x
    add '<python_install_dir>' and '<python_install_dir>\Scripts' to your System Path Variable
2. > pip3 install numpy cython pyqt5 pyopengl
3. Download & install 'Make for Windows' from here: http://gnuwin32.sourceforge.net/packages/make.htm
    add '<make_install_dir>\bin' to System Path
4. Download & install MinGW from here: http://www.mingw.org 
5.  from within MinGW Installation Manager, install
    - mingw32-base
    - mingw32-gcc
    - mingw-developer-toolkit
    - msys-base
6. GIVE UP! :D

Build (native parts)

make

Run

Simulation with GUI

python3 simulation_gui.py

Distributed System

Roles:

• Manager:

  • providing JobQueue and ResultQueue
  • default port: 33333

• Redis Server:

  • acting as key-value store for planet data
  • default port: 6379

• Worker(s):

  • connect to Manager & Redis
  • pull an item off the JobQueue
  • pull planet data from redis
  • calculate result data
  • store result in redis

Worker Setup

• ArchlinuxARM (tested with Raspberry Pi)

  # install dependencies
  sudo pacman -Sy git make gcc python python-numpy cython
  
  # compile cython code
  make

  Default ports:

  • manager: 33333
  • redis: 6379

• Archlinux:

  See Dockerfile

  and

  Makefile

Run distributed system

TODO document starting redis

• Manager

  # launch on default port 33333
  make clmanager

• Workers

  make clworker HOST=<hostname_of_manager>
  
  # or (for non-localhost cluster)
  python3 distributed_worker.py <m_host> <m_port> <r_host> <r_port>