Load test tools for MongoDB in docker containers

Setup runners

# install prerequisite for the load test runner
% pip install -r requirements.txt

# install prerequisite for the ruby runner
% cd lib/ruby_agent
% bundle install

Setup Mongodb docker container cluster

You can choose to test either a Mongo2 cluster or a Mongo3 cluster This will start a Mongodb container and a Graphite container (through which the stats/metrics can be viewed)

for Mongo2

% cd docker/
% docker-compose -f docker-compose-mongo2.6.11.yml up -d

for Mongo3

% cd docker/
% docker-compose -f docker-compose-mongo3.2.8.yml up -d

More docker commands can be found at docker/README.md

Write a test plan

Example can be found at plans/example_plan.yml:

runner:
  language: ruby
  location: ./lib/ruby_agent/

mongodb:
  host: 192.168.99.101
  port: 27017

seed:
  min_id: 1
  max_id: 1000

load_profile:
  write:
    num_processes: 1
    num_requests: 1000
    sleep: 0
  read:
    num_processes: 1
    num_requests: 1000
    sleep: 0
  batchwrite:
    num_processes: 1
    num_requests: 500
    sleep: 0.5
  batchread:
    num_processes: 1
    num_requests: 1000
    sleep: 0.1

Execute the plan

This will execute data seeding and then spawn multiple processes according to the plan against the Mongodb cluster.

% ./load_test_runner.py --plan-file plans/example_plan.yml

View the metrics in Graphite

on Mac

Find the ip address for docker

% echo $DOCKER_HOST
tcp://192.168.99.101:2376

Go to browser: http://192.168.99.101/dashboard

on Linux

Find the ip address for the graphite container

# look for "docker_graphite_1"
% docker network inspect docker_default

Go to browser: http://172.18.0.3/dashboard