Add readinessProbe/minReadySeconds to kube-router

[twz123]

Description

This allows for better feedback of kube-router health via the DaemonSet resource. Without those, it's possible to observe a "healthy" DaemonSet, even if it's not. This affects e.g. rolling updates, and, most notably k0s's own integration tests.

See:

• kube-router failed to start when installation with default settings #4411
• Change kine metrics port from 8080 to 2380 #4421

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update

How Has This Been Tested?

• Manual test
• Auto test added

Checklist:

• My code follows the style guidelines of this project
• My commit messages are signed-off
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• Any dependent changes have been merged and published in downstream modules
• I have checked my code and corrected any misspellings

[jnummelin]

Seems bit excessive? What's the reasoning for such a long delay?

[twz123]

In my experience, liveness probes are only helpful in very few cases. Forcefully restarting a container over and over again is usually not helping much and will just increase churn/load on a cluster that is probably already busy with other things that lead to healthz answering with non-2xx responses. That's why I prefer high timeouts here. Henning wrote a blog post about this back in the day.

[twz123]

The basic difference here is that an app usually reports failures that it might be able to recover from by itself via the readiness endpoint. Restarting the app won't help and might even worse the situation in such a case. Unrecoverable errors should make an app terminate itself. This leaves the liveness probe to detect situations in which the app itself is broken due to things like deadlocks, tight endless loops, blocked on system calls that usually don't block for too long.

[jnummelin]

👍

[k0s-bot]

Successfully created backport PR for release-1.30:

• [Backport release-1.30] Add readinessProbe/minReadySeconds to kube-router #4471