The Crunch.io Dashboard allows users to launch scientific computing clusters in EC2 with ease.
I will be updating these instructions in the near future as well as publicly announcing the opensource release of the core dashboard component.
Steps for running dash on Ubuntu 10.04:
mkvirtualenv --no-site-packages crunchio git clone git://github.com/godber/crunch.io-dashboard.git dash cd dash # Install the system dependencies sudo apt-get install -y python-dev libgmp3-dev libgmp3c2 build-essential # Install the python dependencies pip install -r requirements.txt # Run the following commands and follow the instructions fab setup fab bootstrap
After you have completed the setup steps you can run the app with the following command:
python ./manage.py runserver
And then in another terminal, run the django-task daemon:
python ./manage.py taskd
If you want to deploy this app into production, you can use apache and mod_wsgi and manage the taskd with supervisor. The scripts/ directory contains a django.wsgi file and supervisor configuration file to get you started.
I am trying to establish which testing frameworks I intend to use for this application. At the moment, I have the following two expectations:
- Use doctests to provide functional usage examples is encouraged. But it is not suitible for general testing.
- I may make use of django.test TestCases
- Lettuce tests are the current focus, though that may change.
I have been hoping to use pyccuracy, but I haven't managed to get its basic example working correctly even. I have had no response on their mailing list. I would prefer a BDD style testing framework that actually drives a browser like pycurracy but lettuce is BDD and runs quickly, so perhaps it is suitable for some things.
The following two commands can be used to run tests:
# running the doctests and TestCases ./manage.py test cluster # running the lettuce tests ./manage.py harvest
The dashboard is dependant on a number of open source components and libraries, most notably StarCluster.
Austin