svejnohatomas / poc-artillery.io

A Proof of Concept for using Artillery for backend performance testing, publishing metrics to InfluxDB 2.0 through Telegraf, and displaying them using Grafana.

Proof of Concept - Artillery

Introduction

A Proof of Concept for using Artillery for backend performance testing, publishing metrics to InfluxDB 2.0 through Telegraf, and displaying them using Grafana.

Prerequisite

• Docker
• Node.js (tested with Node.js 20 LTS)

Run and setup testing infrastructure

1. Run Docker Compose: docker-compose up -d

   • This will run InfluxDB 2.0, Telegraf, Grafana and a Weather Forecast API

2. Go to your Grafana and log in

   • username: admin
   • password: admin

3. Connect Grafana to Influx DB

   • Go to Add data source, select InfluxDB and configure the following

     Query Language: Flux # InfluxDB 2.0
     HTTP:
       URL: http://influxdb2:8086 # InfluxDB address inside Docker
     Auth:
       Basic auth: false # Turn off Basic Auth
     InfluxDB Details:
       Organization: poc-organization
       Token: my-super-secret-auth-token # You should issue a new Read-Only access token for Grafana in your InfluxDB
       Default Bucket: poc-artillery.io

4. Create a new Grafana Dashboard (you can find the finished dashboard here)

   • Go to Dashboards and click New Dashboard
   • Click Add an empty panel

     • Responses

       • Panel (Settings)

         • Panel title: Responses

       • Query

         from(bucket: "poc-artillery.io")
           |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
           |> filter(fn: (r) => r["testId"] == "weather-forecast")
           |> filter(fn: (r) =>
                 r["_measurement"] =~ /artillery_http_codes_1[0-9]{2}/ or
                 r["_measurement"] =~ /artillery_http_codes_2[0-9]{2}/ or
                 r["_measurement"] =~ /artillery_http_codes_3[0-9]{2}/ or
                 r["_measurement"] =~ /artillery_http_codes_4[0-9]{2}/ or
                 r["_measurement"] =~ /artillery_http_codes_5[0-9]{2}/)
           |> aggregateWindow(every: v.windowPeriod, fn: mean, createEmpty: false)
           |> yield(name: "mean")
         

       • Overrides (Fields with name matching regex)

         • 200 Responses
           • RegEx: artillery_http_codes_200.+
           • Display name: 200
         • 400 Responses
           • RegEx: artillery_http_codes_400.+
           • Display name: 400
         • 404 Responses
           • RegEx: artillery_http_codes_404.+
           • Display name: 404

     • HTTP Response Time (Latency)

       • Panel (Settings)

         • Panel title: HTTP Response Time (Latency)

       • Field (Standard options)

         • Unit: milliseconds (ms) (under Time)

       • Query

         from(bucket: "poc-artillery.io")
           |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
           |> filter(fn: (r) => r["testId"] == "weather-forecast")
           |> filter(fn: (r) =>
                 r["_measurement"] == "artillery_http_response_time_max" or
                 r["_measurement"] == "artillery_http_response_time_median" or
                 r["_measurement"] == "artillery_http_response_time_min" or
                 r["_measurement"] == "artillery_http_response_time_p95" or
                 r["_measurement"] == "artillery_http_response_time_p99")
           |> aggregateWindow(every: v.windowPeriod, fn: mean, createEmpty: false)
           |> yield(name: "mean")
         

       • Overrides (Fields with name matching regex)

         • Maximum latency
           • RegEx: artillery_http_response_time_max.+
           • Display name: Maximum
         • Median latency
           • RegEx: artillery_http_response_time_median.+
           • Display name: Median
         • Minimum latency
           • RegEx: artillery_http_response_time_min.+
           • Display name: Minimum
         • 95th percentile latency
           • RegEx: artillery_http_response_time_p95.+
           • Display name: 95th percentile latency
         • 99th percentile latency
           • RegEx: artillery_http_response_time_p99.+
           • Display name: 99th percentile latency

     • Traffic

       • Panel (Settings)

         • Panel title: Traffic

       • Field (Standard options)

         • Unit: requests/sec (rps) (under Throughput)

       • Query

         from(bucket: "poc-artillery.io")
           |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
           |> filter(fn: (r) => r["testId"] == "weather-forecast")
           |> filter(fn: (r) => r["_measurement"] == "artillery_http_request_rate")
           |> aggregateWindow(every: v.windowPeriod, fn: mean, createEmpty: false)
           |> yield(name: "mean")
         

       • Overrides (Fields with name matching regex)

         • Maximum latency
           • RegEx: value.+
           • Display name: Number of requests

Run Tests

1. Go to test folder: cd .\test\performance-testing\
2. Install dependencies: npm install
3. Run tests: npm run weather-forecast

You should now see your Artillery output in Grafana.
If you can't see anything in Grafana, check InfluxDB bucket (Data/Buckets/poc-artillery.io).

Authors

• Tomas Svejnoha
  • GitHub: svejnohatomas
  • LinkedIn: Tomáš Švejnoha