v0.11.6 Testing checklist

[jorgeepc]

Testing Tracetest

This is the entire checklist on what we should do to assert that Tracetest is working fine on each version release. On each version release, we can copy the contents of this template checklist and open a Github Discussion to start each test.

On Tracetest, we work in two ways to test the entire system and guarantee that everything is working fine. One is through automatic tests, where we run unit, integration and end-to-end tests on Tracetest CLI, Web UI and Server.

Another source of tests is the manual tests that we execute on each release, simulating users, and checking if everything is ok on these tests.

Manual Tests

On our manual tests, we aim to do some sanity checks to see if the main features are working as expected. Usually, we run it on each Tracetest release.

Tests to validate on Release Candidate

• Test server installation via CLI

  • Docker Compose and no demo API
  • Docker Compose and demo API
  • Kubernetes and no demo API
  • Kubernetes and demo API
  • Verify that Tracetest is not sending analytics if you have the TRACETEST_DEV env for docker install
  • Verify that Tracetest is not sending analytics if you have the TRACETEST_DEV env for k8s install

• Test DataStore provisioning and basic test

  • Open WebUI and go to /settings page. The provisioned Data Store should be selected
  • Create a test on WebUI that calls a demo API (like Pokeshop or Open Telemetry Store). This test should fetch traces correctly and run without errors

• Test Tracetest examples

  • Jaeger example
  • Tempo example
  • New Relic example
  • Lightstep example
  • OpenSearch example
  • Elastic APM example
  • Datadog example
  • Amazon X-Ray example

• Test specific features added/changed on this release:

  • Validate new CLI datastore resource: apply datastore, get datastore, delete datastore
  • Validate CLI config resource: apply config, get config
  • Validate CLI pollingprofile resource: apply pollingprofile, get pollingprofile
  • Validate CLI demo resource: apply demo, get demo, delete demo, list demo
  • Validate webUI datastore CRUD
  • Test HTTP OTLP server
  • Create a test with an URL without scheme, it should default to http

Tests to validate on Final Release

• Pipeline

  • Check if our release pipeline on Release Tracetest workflow on Github Actions worked correctly

• Test CLI updates

  • MacOS via homebrew
  • MacOS via curl script
  • Windows via chocolatey
  • Windows via manual download

• Docs

  • Double check Detailed installation docs and see if everything is documented correctly

Automatic Tests

Today Tracetest has 3 main components: a WebUI, a CLI and a Server.

Web UI

• Unit tests: Run by executing npm test on ./web folder
• End-to-end tests: Run using cypress against a temporary Tracetest created on Kubernetes.

CLI

• Unit tests: Run by executing make test on ./cli folder

Server

• Unit tests: Run by executing make test on ./server folder
• Integration tests: Run along with some unit tests running make test on ./server folder, but also done in a matrix test on Github actions, by executing the test-examples action.
• End-to-end tests: Run using Tracetest to test itself. We deploy two instances of Tracetest and use one to test API calls to another on Github actions, by executing the trace-testing action.

[danielbdias]

Found a bug when running Tempo Cloud example. Will open a PR to solve that.

[danielbdias]

Solved on #2444

[jorgeepc]

Found a new issue with data store types for opensearch, elasticapm, signalfx and newrelic. PR to fix it here #2454

[danielbdias]

Found a small configuration error on Kubernetes and demo API that makes PokeAPI to not report the traces correctly, I'm reopen #2143 and will tackle this problem there.