ardere

AWS Serverless Service for Load-Testing

ardere runs as a serverless service using AWS to orchestrate load-tests consisting of docker container configurations arranged as test plans.

Installation

Pre-requisite: installation requires node > v6

To deploy ardere to your AWS account, you will need a fairly recent install of Node, then install the Node packages required:

$ npm install

You will need to ensure your have AWS access and secret keys configured for serverless:

$ sls config

To deploy the ardere lambda's and required AWS stack:

$ sls deploy

Then you can deploy the ardere Step Function:

$ sls deploy stepf

Developing

ardere is written in Python and deployed via serverless to AWS. To an extent testing it on AWS is the most reliable indicator it works as intended. However, there are sets of tests that ensure the Python code is valid and works with arguments as intended that may be run locally.

Create a Python virtualenv, and install the test requirements:

$ virtualenv ardenv
$ source ardenv/bin/activate
$ pip install -r test-requirements.txt

The tests can now be run with nose:

$ nosetests

Note that you cannot run the sls deploy while the virtualenv is active due to how the serverless Python requirements plugin operates.

Run Test

1. Login to AWS console (mozilla-services use: stage)
2. Go to Step Functions > Dashboard
3. Select your state machine (mozilla-services use: "ardere-dev-ardere")
4. Click on "New Execution" button
5. Paste your json config into text area (example: mozilla-services/screenshots-loadtests /ardere.json)
6. Optional: Assign a name to your execution
7. Click on "Start Execution"
8. Monitor execution in Dashboard
9. Test load should be visible in DataDog, NewRelic, etc.

Monitoring

Metrics Node Monitoring (Grafana)

1. ssh -L 3000:<ip_metrics_node>:3000 <ip_bastion_host>
2. open local browser to http://localhost:3000
3. login using credentials specified in your ardere (JSON) config file