This is your first blog post. Replace this content with your actual article.
Introduction
Write your blog post content here in Markdown. Claude Code will create these files for you when you share your notes.
Vertical Pod Autoscaler
The VPA adjusts CPU and memory requests automatically based on observed usage. Start in recommendation mode before enabling auto-mode.
kubectl top nodes
kubectl top pods --all-namespaces --sort-by=memorySpot Instance Strategy
Running non-critical workloads on spot instances can cut compute costs by 60–80%. Use node affinity rules and Pod Disruption Budgets to handle interruptions gracefully.
Right-sizing with Goldilocks
The Goldilocks tool runs VPA in recommendation mode across all namespaces and exposes a dashboard showing suggested requests and limits.
Summary
- Enable VPA in recommendation mode first, then auto
- Use Spot/Preemptible nodes for stateless workloads
- Set resource requests and limits on every pod
- Use Horizontal Pod Autoscaler with custom metrics
- Review Cluster Autoscaler scale-down settings