helm-node-problem-detector

A Crossplane Configuration package that installs the node-problem-detector Helm chart with a minimal, stable interface.

Overview

helm-node-problem-detector renders a single Helm release for node-problem-detector. It exposes only the inputs needed for chart values, namespace, and release name, keeping the interface stable while allowing full Helm overrides.

Node Problem Detector is a DaemonSet that detects node problems and reports them to the API server. It can detect issues like kernel deadlocks, corrupted filesystems, and container runtime problems.

Features

• Minimal Helm interface: values and overrideAllValues with stable defaults
• Predictable naming: defaults to node-problem-detector in the kube-system namespace
• GitOps friendly: ships a .gitops/ deploy chart

Prerequisites

• Crossplane installed in the cluster
• Crossplane providers:
  • provider-helm (>=v1.0.6)
• Crossplane function:
  • function-auto-ready (>=v0.6.0)

Quick Start

apiVersion: pkg.crossplane.io/v1
kind: Configuration
metadata:
  name: helm-node-problem-detector
spec:
  package: ghcr.io/hops-ops/helm-node-problem-detector:latest

apiVersion: helm.hops.ops.com.ai/v1alpha1
kind: NodeProblemDetector
metadata:
  name: node-problem-detector
  namespace: example-env
spec:
  clusterName: example-cluster
  values:
    metrics:
      enabled: true
      serviceMonitor:
        enabled: true

Detected Problems

Node Problem Detector can detect:

• Kernel deadlocks
• Corrupted filesystems
• Container runtime issues
• Hardware problems
• Kernel OOM events
• And more through custom plugins

Development

make render
make validate
make test