Add component node-problem-detector

[knrt10]

This adds component node-problem-detector. For more information
refer to https://github.com/kubernetes/node-problem-detector.

closes: #39
Signed-off-by: knrt10 [email protected]

TODO

• Add tests
• Add component to CI

How to use

Pass the following in your cluster config. To use service_monitor feature, make sure Prometheus Operator component is installed

component "node-problem-detector" {
  service_monitor = true
}

[invidian]

Some nits, otherwise LGTM.

I wonder if we could have some e2e test to ensure that this detector is doing it's thing and has privileges to modify Node object etc.

[knrt10]

Updated the PR, writing e2e test, to test the working of component

[invidian]

I feel like we could squash almost everything into single commit. Maybe we could keep asset import in a separate one, as it's a lot of files.

[invidian]

And do we have a way of testing custom monitors? Did you test it @knrt10 ?

[surajssd]

How is this integrating with the prometheus stack?

I am trying to understand once this is installed now what? How does the user get any alerts?

[knrt10]

@surajssd the component has Prometheus exporter. I will test it with our Prometheus component and update the PR

[knrt10]

@surajssd @invidian I have added a new knob service_monitor. I have tested the working of metrics, it works fine. So the use of components will be like this

• User installs the component with service_monitor enabled then they can scrape metrics from prometheus

[invidian]

We can test that this metric can be scraped via e2e tests then.

[surajssd]

LGTM :-)