Documentation

Vigil is a Kubernetes controller that prevents workload scheduling on new nodes until all expected DaemonSet pods are running and Ready. It uses Karpenter’s startupTaints mechanism to gate node readiness, eliminating the scheduler cache race condition that causes pod rejections during node startup.

Getting Started

Install Vigil in your cluster and configure Karpenter startup taints. The Getting Started guide covers prerequisites, Helm installation, and verification.

Concepts

Understand how Vigil works:

• Architecture — Controller design, informers, reconciliation loop
• DaemonSet Discovery — How Vigil determines which DaemonSets should run on each node

Reference

• Configuration — All configuration options
• Metrics — Prometheus metrics catalog
• Helm Chart — Helm values reference