Create e2e tests for all exposed Prometheus metrics

[zroubalik]

Proposal

At the moment KEDA exposes following metrics:

• keda_scaler_active
• keda_scaler_metrics_value
• keda_scaler_metrics_latency
• keda_scaler_errors
• keda_scaler_error_totals
• keda_scaled_object_errors
• keda_resource_totals
• keda_trigger_totals

We should cover all of these metrics by e2e tests:

• test that metrics are exposed
• test that metrics are correctly calculated (ie. we should test that metrics related to errros are correctly incremented in case of an error in a scaler/scaledobject, etc.).

Exisiting Prometheus Metrics e2e tests: https://github.com/kedacore/keda/blob/main/tests/internals/prometheus_metrics/prometheus_metrics_test.go

Docs: https://keda.sh/docs/2.10/operate/prometheus/

[Parthiba-Hazra]

hey, is this issue already worked on ? if not then I want to give a try... as I'm new in open source.

[JorTurFer]

Hey @Parthiba-Hazra
There isn't anybody working on this AFAIK, I assign the issue to you 😄
Thanks for the help!

[Parthiba-Hazra]

Hey @JorTurFer thank you for the response .
As a new folk how should I approach this issue, anything that could help me..

[zroubalik]

@Parthiba-Hazra there's a doc around e2e tests: https://github.com/kedacore/keda/tree/main/tests#readme

Existing Prometheus metrics related tests are here:
https://github.com/kedacore/keda/blob/main/tests/internals/prometheus_metrics/prometheus_metrics_test.go

In the issue description you can see missing tests, so you need to create test cases for them.

[Parthiba-Hazra]

got it 👍

[Parthiba-Hazra]

Hey @zroubalik I need some help --
for the keda_scaler_errors test case, first I have to apply a scaled_object that will cause error with Prometheus , then get the operator metrics and check that keda operator show the errors...
is that right?

[zroubalik]

@Parthiba-Hazra yeah, check that Prometheus metrics show the error.

[Parthiba-Hazra]

hey @zroubalik I created test cases that asserts that applying wrong scaled object increase the counter-value of keda_scaler_errors , keda_scaler_error_totals and keda_scaled_object_errors metrics from 0 to any positive number .. is there anything that I should check also ?