Kubernetes Orchestration - Pods, Deployments, Services & Ingress

etcd is a strongly consistent, distributed key-value store that serves as Kubernetes' sole source of truth. It stores all cluster state, resource specifications, and runtime metadata. If etcd loses data without a backup, the cluster state is lost.

The kube-scheduler continuously monitors unassigned Pods and evaluates nodes using a two-step process:
1. Filtering (Predicates): Excludes nodes that fail resource requests, taints/tolerations, or affinity rules.
2. Scoring (Priorities): Ranks remaining nodes to select the optimal node for Pod execution.

The kube-apiserver serves as the front door and central management hub for the Kubernetes control plane. It exposes the REST API, validates and processes object specifications (Pods, Services, Deployments), and writes state data to etcd.

The kube-controller-manager runs continuous control loops that drive the cluster from its current state toward the desired state. Key controllers include the Node Controller, ReplicaSet Controller, and Endpoints Controller.

Init Containers are specialized containers that run sequentially to completion before app containers start in a Pod. They are typically used for setup tasks like running database migrations, fetching secrets, or waiting for external dependencies to become available.

A Deployment is a higher-level controller that manages declarative updates for application Pods. It automatically manages underlying ReplicaSet objects, which in turn scale and maintain the exact target count of identical running Pod instances.