A set of tasks for use with Fabric and EC2.
Now available on PyPi at http://pypi.python.org/pypi/cloth. Installation is therefore as simple as:
pip install cloth
Export your EC2 credntials as environment variables.
export AWS_EC2_REGION=eu-west-1 export AWS_ACCESS_KEY_ID=<your-access-key> export AWS_SECRET_ACCESS_KEY=<your-secret-key>
To use just import some or all of the tasks into your fabric file. Or create a blank fabfile.py with the following contents.
#! /usr/bin/env python from cloth.tasks import *
This will give you a good few commands.
⚡ fab -l Available commands: all All nodes free Show memory stats list List EC2 name and public and private ip address nodes Select nodes based on a regular expression preview Preview nodes production Production nodes updates Show package counts needing updates upgrade Upgrade packages with apt-get uptime Show uptime and load
Of most interest should be the 'all' and 'nodes' tasks. These allow you to load EC2 instances for further command running.
⚡ fab all list
The above should list all of your EC2 instances including the name and public and private ip addresses.
⚡ fab nodes:"^production.*" list
The above should list all of your EC2 instances that start with 'production'. This takes a regex as the argument so you can get whatever instances you like.
⚡ fab all uptime
As an example of running a command on a set of EC2 instances try the above. This should show the uptime and load averages for all your EC2 instances. Use -P as well to have that happen in parallel.
I generally use a convention for the names of my EC2 machines, in particular:
The production and preview tasks simple filter for those with a platform value of production or preview. More interesting is that roles are being set based on the second part of the name. For instance if I have a set of instances called:
I could write a task like so:
@task @roles('backend') def passenger(): "Show details about passenger performance" sudo('passenger-memory-stats') sudo('passenger-status')
The run that task with:
⚡ fab all passenger
That task would only be run on the three backend instances.