apistress

This is very simple stress testing tool for API based on vegeta

Motivation

Sometimes you want to check SLA of your API automatically. There are many tools for profiling or stressing your software, but all of them have the enormous complexity or analyze results by third party tools. This project helps you do it easy and in a straightforward way. One configuration file with a simple structure is needed. In configuration file you need to declare the target, attack time, request frequency and (this is the most important part) SLA for target service (99th latency percentile and percentage of successful requests). And if any test is failed, the program will print report and exit with error code. It gives simple integration with different continuous delivery tools.

Usage

Create file with name config.json and folowing structure:

{
  "baseUrl": "http://localhost:8080",               //base url for targets
  "tests": [                                        //array of tests
    {
      "rps": 100,                                   //target request per second
      "duration": 1,                                //test duration
      "target": {
        "method": "POST",                           //http method for target url
        "path": "/test",                            //relative path
        "headers": [                                //optional
          {
            "name": "X-Request-Id",                 //http header name
            "value": "12345"                        //http header value
          }
        ],
        "body": "ewoic2F5IiA6ICJoZWxsbyIKfQ=="      //base64 endoded request body (optional)
      },
      "sla": {
        "latency": 150,                             //99 percenitle latency
        "successRate": 99.9                         //percentage of successful requests (2xx http code is returned)
      }
    }
  ]
}

Then run docker container:

docker run --rm --net=host \
   -v /path/to/folder/with/config:/data \
   aatarasoff/apistress

or with overriden baseUrl config property:

docker run --rm --net=host \
   -v /path/to/folder/with/config:/data \
   aatarasoff/apistress apistress \
   -baseUrl http://custom.server:8080

If stdin input is required, use -config=stdin flag:

cat config.json | docker run --rm --net=host \
   aatarasoff/apistress apistress \
   -config=stdin

Also it is possible to define own config file name and path:

docker run --rm --net=host \
   -v /path/to/folder/with/config:/data \
   aatarasoff/apistress apistress \
   -config=/path/to/folder/with/config/filename.json

For each test program prints metrics into stdout:

Requests      [total, rate]            10, 11.11
Duration      [total, attack, wait]    1.007898226s, 899.953255ms, 107.944971ms
Latencies     [mean, 50, 95, 99, max]  108.246432ms, 107.608893ms, 109.534083ms, 109.534083ms, 112.276495ms
Bytes In      [total, mean]            5750, 575.00
Bytes Out     [total, mean]            0, 0.00
Success       [ratio]                  100.00%
Status Codes  [code:count]             200:10
Error Set:

and returns exit code that you may check with following command:

echo $?       //0 - ok, 1 - sla error

Usage without Docker or developing

You need install and setup golang 1.6 or above with following instructions. Then run:

go get
go install github.com/aatarasoff/apistress