kubespray-ha-configuration
Installation
SKILL.md
Kubespray HA Configuration
Overview
High availability in Kubernetes requires redundant control plane nodes and etcd members. Kubespray automates HA setup but understanding the architecture is essential for proper sizing and troubleshooting.
Core principle: API servers are active-active (all handle requests). Controller-manager and scheduler use leader election (one active, others standby). etcd requires quorum (majority must agree).
When to Use
- Deploying production clusters requiring uptime guarantees
- Sizing etcd clusters (why odd numbers matter)
- Configuring load balancing for API server
- Understanding failover behavior
Not for: Single-node deployments (use kubespray-deployment), troubleshooting HA failures (use kubespray-troubleshooting)