docs

Get started

Welcome
Installation
Running k6
Results output
k6 resources

Using k6

HTTP Requests
Metrics
Checks
Thresholds
Options
Test lifecycle
Modules
Tags and Groups
Cookies
Protocols
Environment variables
Execution context variables
Scenarios

Using k6 browser

Overview
Running browser tests
Browser metrics
Migrating to k6 v0.46
Recommended practices

Results output

Overview
End of test
Real time
Grafana dashboards

Test authoring

Test builder
Create tests from recordings

Test types

Load test types
Smoke testing
Load testing
Stress testing
Soak testing
Spike testing
Breakpoint testing

Testing guides

API load testing
Automated performance testing
Load testing websites
Running large tests
Running distributed tests

Misc

Usage collection
IntelliSense
Fine tuning OS
Archive Command
Release notes
Glossary
k6.ioapp.k6.io
⚠️ This documentation is outdated. Please visit grafana.com for the latest k6 documentation.📚
  • Using k6
  • Workaround to calculate iteration_duration

Workaround to calculate iteration_duration

suggest edits

A common requested case is to track the iteration_duration metric without including time spent for setup and teardown functions. This feature is not yet available but a threshold on iteration_duration or any pre-existing metrics can be used as a workaround.

It's based on the concept of creating thresholds for sub-metrics created by tags for the required scope and setting the criteria that always pass. It works with any enabled tags that already works with threshold, for example:

  • iteration_duration{scenario:default} generates a sub-metric collecting samples only for the default scenario's iteration. scenario:default is used because that's the internal k6 scenario name when we haven't specified options. scenarios` explicitly and are just using the execution shortcuts instead.
  • iteration_duration{group:::setup} or iteration_duration{group:::teardown} create sub-metrics collecting the duration only for setup and teardown. k6 implicitly creates groups for setup and teardown, and :: is the group separator.
  • http_req_duration{scenario:default} can be useful as well for isolating executed long-running requests.
1import { sleep } from 'k6';
2import http from 'k6/http';
3

4export const options = {
5  vus: 1,
6  iterations: 1,
7  thresholds: {
8    'iteration_duration{scenario:default}': [`max>=0`],
9    'iteration_duration{group:::setup}': [`max>=0`],
10    'iteration_duration{group:::teardown}': [`max>=0`],
11    'http_req_duration{scenario:default}': [`max>=0`],
12  },
13};
14

15export function setup() {
16  http.get('https://httpbin.test.k6.io/delay/5');
17}
18

19export default function () {
20  http.get('http://test.k6.io/?where=default');
21  sleep(0.5);
22}
23

24export function teardown() {
25  http.get('https://httpbin.test.k6.io/delay/3');
26  sleep(5);
27}

Dedicated sub-metrics have been generated collecting samples only for the scope defined by thresholds:

     http_req_duration..............: avg=1.48s    min=101.95ms med=148.4ms  max=5.22s    p(90)=4.21s    p(95)=4.71s
       { expected_response:true }...: avg=1.48s    min=101.95ms med=148.4ms  max=5.22s    p(90)=4.21s    p(95)=4.71s
     ✓ { scenario:default }.........: avg=148.4ms  min=103.1ms  med=148.4ms  max=193.7ms  p(90)=184.64ms p(95)=189.17ms


     iteration_duration.............: avg=5.51s    min=1.61s    med=6.13s    max=8.81s    p(90)=8.27s    p(95)=8.54s
     ✓ { group:::setup }............: avg=6.13s    min=6.13s    med=6.13s    max=6.13s    p(90)=6.13s    p(95)=6.13s
     ✓ { group:::teardown }.........: avg=8.81s    min=8.81s    med=8.81s    max=8.81s    p(90)=8.81s    p(95)=8.81s
     ✓ { scenario:default }.........: avg=1.61s    min=1.61s    med=1.61s    max=1.61s    p(90)=1.61s    p(95)=1.61s