charts/vertical-pod-autoscaler: Add PodDisruptionBudgets and ServiceMonitor support #253
Description
Summary
Enhance the vertical-pod-autoscaler chart production readiness by adding support for PodDisruptionBudgets and ServiceMonitor resources.
Background
The vertical-pod-autoscaler chart (introduced in PR #250) mentioned these features as future todos. These additions would significantly improve the chart's operational capabilities for production deployments.
Proposed Features
1. PodDisruptionBudgets (PDB)
Add optional PodDisruptionBudgets for each VPA component to ensure availability during cluster maintenance:
Benefits:
- Prevents all instances of a component from being evicted simultaneously
- Ensures VPA functionality during node maintenance or upgrades
- Critical for production stability
Implementation:
- Optional PDB templates for recommender, updater, and admission-controller
- Configurable minAvailable or maxUnavailable settings
- Default to disabled to maintain backwards compatibility
2. ServiceMonitor Support
Add optional ServiceMonitor resources for Prometheus integration:
Benefits:
- Enables automatic discovery of VPA metrics by Prometheus
- Provides observability into VPA component health and performance
- Standard monitoring integration for Kubernetes clusters
Implementation:
- Optional ServiceMonitor templates for each component
- Configurable scrape intervals and paths
- Support for additional labels and annotations
- Integration with existing Service resources
Implementation Notes
- All features should be optional and disabled by default
- Follow existing chart patterns and naming conventions
- Update README.md with configuration examples
- Consider component-specific requirements (e.g., admission-controller typically runs single instance)
Related
- Original mention in PR charts/vertical-pod-autoscaler new chart (closes #249) #250 summary
- Follows production readiness best practices for Helm charts in this repository