Payments Demo Load tests

This k6-based project is built with npm accordingly to this documentation.

Tests are developed for this repository performance testing.

System requirements

• node ^20.*
• npm ^10.*

Tests configuration

Instead of sending many flags to k6 run, we have all the load run configs right in options constant per test suite. Default configuration is to run load for 20 virtual users for 5 or 10 minutes (depending on test type) as ramp up, then stay with this amount of users for the next 10 minutes and finally ramp-down to 0 users in last 5 minutes of run. We also have thresholds check that 90% of calls' responses took less than 1 sec, and we had less than 3% failed responses from total amount.

Tests execution

Reach out to official k6 installation docs.

First, you need to have payments-demo-quarkus-app running locally.

To run test with k6 installation:

k6 run payments.test.js

Load test results to Grafana

There is a nice article about how to load tests results to influxdb and consume it in grafana. Unfortunately, it works fine only in some specific platforms and very specific versions of k6, grafana and influxdb.

InfluxDB 2

In case of using OS X or any other platform where influxdb 1.8 doesn't work perfectly fine, we might need to use influxdb version 2+ that is not currently supported with k6 tool. Then we need to compile our tests to binary with a special took xk6. Reach out to documentation for installation details. Next we should run a command to generate k6 binary with in the project: xk6 build --with github.com/grafana/xk6-output-influxdb.

Now we can run our tests using this binary and expose results to influxdb. First, run a container with influxdb instance:

docker run -d -p 8086:8086 \
  --env "DOCKER_INFLUXDB_INIT_MODE=setup" \
  --env "DOCKER_INFLUXDB_INIT_USERNAME=admin" \
  --env "DOCKER_INFLUXDB_INIT_PASSWORD=password1" \
  --env "DOCKER_INFLUXDB_INIT_ORG=k6io" \
  --env "DOCKER_INFLUXDB_INIT_BUCKET=load_tests" \
  --env "DOCKER_INFLUXDB_INIT_ADMIN_TOKEN=secret_token" \
  influxdb:2.5.1

Now you can run tests and expose results to docker container:

K6_INFLUXDB_ORGANIZATION=k6io K6_INFLUXDB_TOKEN=secret_token ./k6 run -o xk6-influxdb=http://127.0.0.1:8086/load_tests tests/cache.test.js

To check that results were uploaded correctly, you can go to http://localhost:8086, then navigate to Data Explorer and observe load_tests bucket data.

Grafana

We can start an instance of grafana as a standalone service (brew install grafana && brew services start grafana for OS X) or as a container:

docker run -d -p 3000:3000 \
  --env "GF_AUTH_ANONYMOUS_ORG_ROLE=Admin" \
  --env "GF_AUTH_ANONYMOUS_ENABLED=true" \
  --env "GF_AUTH_BASIC_ENABLED=false" \
  -v $PWD/grafana:/etc/grafana/provisioning/ \
  grafana/grafana:10.1.2

Then navigate to http://localhost:3000 and login with admin / admin. We need to configure influxdb source and the easiest way is to go Data sources and select InfluxDB, then just fill in URL with http://localhost:8086 and add a header Authorization with value Token secret_token. Go to details section and fill in Database with load_tests and then click on Save & test button.

We're ready to configure dashboard, so you can simply import existing k6 dashboard template and choose your configured influxdb source as per documentation.

Testkube

We can run load tests in kubernetes cluster to make sure we only measure application's load and not depend on network lag. It's possible via deploying Testkube helm chart into existing kubernetes cluster and configure it to run these load tests.

Installing chart and running tests

• run

chmod +x ./scripts/*
./scripts/start-helm.sh # password will be asked for minukube tunnel - keep terminal opened
./scripts/grafana-dashboard.sh # in a new terminal

• open testkube dashboard
• navigate to payments-test
• click Run now
• open running test execution to see real-time results

Observing Grafana results

• open grafana
• navigate to dashboards
• select pre-configured k6 Load Testing Results dashboard
• once you started tests, results will be automatically shown
• you can change filters in top right corner to Last 5 minutes to better see the output

Removing chart

chmod +x ./scripts/stop-helm.sh && ./scripts/stop-helm.sh