flypenguin/test repository overview

⁠docker-loadummy

A simple little Flask app which computes pi to the nth number to simulate CPU load to test autoscaling on the cloud provider.

⁠Testing the flask app locally

python flask_app.py

(of course flask has to be installed, so you should do it in a virtualenv)

⁠Environment variables

Displayed are the default values

COLOR=white         # a color name (see below), a color hex code, or "`random`"
STARTUP_TIME=0.0    # sleep that many seconds before flask starts
FLASK_PORT=80       # port flask listens on ()
FLASK_DEBUG=0       # whether flask is put in debug mode
FLASK_THREADED=1    # whether flask is being run multi-threaded
LOADUMMY_NAME=      # optional. set a name, useful for clusters - see below
LOADUMMY_NEXT=      # optional. documentation for this - see below

⁠Data encoding

The app will return HTML data (Content-type: text/html), unless otherwise specified in the Accept: header. If it finds the string yaml in that header, YAML data is returned (Content-type: application/x-yaml, like rails), if it finds the string json it returns JSON data (Content-type: application/json).

⁠Endpoints

⁠Using the /distrib endpoint

Originally used to simulate load. Never really tried, but should work.

To use this endpoint variable LOADUMMY_DISTRIB must be set: LOADUMMY_DISTRIB=https?://<next_host>[:<nextport>]

Now, if you call http://host:port/distrib/5/250 now, what happens then is ...

• it will create NUM requests for calculation of pi with
• (on average) DIGITS digits.

Or better: if you call it with /5/250, it might query those urls:

• http://next_host:nextport/pi/267 (1st request)
• http://next_host:nextport/pi/229 (2nd request)
• http://next_host:nextport/pi/244 (...)
• http://next_host:nextport/pi/231
• http://next_host:nextport/pi/276 (5th request)

... where the numbers are made up randomly around the 250 digits marker (plus minus 10%).

⁠Valid color names