Load tests for Rocket.Chat

This repository implements a load test for Rocket.Chat.

Required Software:

• apg password generator used for generating random usernames
• node
• python-pip – apt-get install python-pip or brew install python@2
• numpy - pip install numpy (for the outcome parser)

Instruction

Install required packages

npm install

Generate random users

./message_listener/lib/generate-users.sh <usercount>

generates two files, user_insert_script.js for inserting into mongo and userdata.js to have the usernames in the script - every testuser has the PW:test123

Import Users

Execute newly generated script to import the users on the mongodb shell -

mongo <serverurl>/<database> user_insert_script.js

Example: mongo 127.0.0.1/meteor ./message_listener/lib/user_insert_script.js

Execute the test

Execute the message listener testcase with the following instructions:

Synopsis

Usage: apps.js OFFSET [options]
    --help, -h
    --message-interval VAL, -i VAL
    --message-interval-jitter VAL, -j VAL
    --message-count VAL, -n VAL
    --server-url VAL, -s VAL
    --receiver-additional-waiting-time VAL, -w VAL
    --user-count VAL, -u VAL

Options

OFFSET:
    User offset (lib/userdata.js), useful when running multiple instances with the same dataset (offset by N).

--message-interval, -i:
    Interval between messages in seconds. Default: 1

--message-interval-jitter, -j:
    Random jitter to add/substract from the message interval. Default: 0.5

--message-count, -n:
    How many messages the sender should send. Default: 10

--server-url, -s:
    RocketChat server url. Default: https://open.rocket.chat/

--receiver-additional-waiting-time, -w:
    Time in seconds to wait for messages after the sender has stopped. Default: 1

--user-count, -u:
    How many clients should be started. Must be a multiple of 2 (sender/receiver). Default: 2

Example

node message_listener/app.js 0 -i 0.1 -j 0.05 -n 10 -u 100 -w 5

Start 100 clients (50 receiver, 50 sender). The sender posts 10 messages at an interval of 0.1 ± 0.05 seconds. The receiver will wait up to 5 seconds for pending messages.

Results

The results can be piped to a file an parsed via the parse_outcome.py

./parse_outcome.py <filename>