Kubernetes controller
A program inside or alongside the cluster that watches one kind of object and keeps acting until reality matches what that object asks for.
In more detail
A Deployment controller keeps the right ReplicaSet, a ReplicaSet controller keeps the right number of pods, and a Job controller keeps retrying until enough pods succeed. A bare Pod has no controller, so nothing replaces it if it dies. Controllers only talk to the API server, never directly to each other.
Related terms
- reconciliation — The repeated check-and-fix loop in which a program compares desired state with observed state and changes the world to match.
- Argo CD — A GitOps tool for Kubernetes that continuously syncs Applications described in Git into clusters.
- Flux — A GitOps tool for Kubernetes built from small controllers that fetch sources and apply them to the cluster.
- desired state — A description of how a machine should look when the work is finished, rather than a list of steps to get there.
- orchestration — Automatically deploying, scaling, restarting and moving containers across a group of machines.
- GitOps — Operating a system by keeping its desired state in Git and letting a controller continuously make the live system match it.
- kube-controller-manager — The control plane component that runs the loops keeping the cluster's actual state matching the desired state.
- Deployment — The Kubernetes object that keeps a chosen number of identical pods running and replaces them gradually when you change their template.
- ReplicaSet — The Kubernetes object that keeps a given number of pod copies running; Deployments manage ReplicaSets for you.
Taught in
- The Push Model vs. The Pull Model — Continuous Integration & Continuous Delivery Pipelines
- What is GitOps? — Continuous Integration & Continuous Delivery Pipelines
- Argo CD Architecture and Applications — Continuous Integration & Continuous Delivery Pipelines
- Flux as the Alternative — Continuous Integration & Continuous Delivery Pipelines
- Containers And Orchestration Context — DevOps & Site Reliability Engineering
- Pipelines And GitOps Context — DevOps & Site Reliability Engineering
- From Docker To Desired State — Kubernetes: Production-Grade Container Orchestration
- Control Plane: etcd, API Server, Scheduler, Controllers — Kubernetes: Production-Grade Container Orchestration
- Namespaces, Labels And Declarative Manifests — Kubernetes: Production-Grade Container Orchestration
- Deployments, ReplicaSets, Rollouts And Rollback — Kubernetes: Production-Grade Container Orchestration
- Jobs And CronJobs — Kubernetes: Production-Grade Container Orchestration
- Troubleshooting: ImagePullBackOff And Failed Rollouts — Kubernetes: Production-Grade Container Orchestration
- NodePort, LoadBalancer And ExternalName — Kubernetes: Production-Grade Container Orchestration
- Ingress: Controllers, Hosts, Paths And TLS — Kubernetes: Production-Grade Container Orchestration