Kubernetes · The fleet
Captain's dictionary
Every word from the ports, in one plain sentence, with the lesson that explains it.
31 words
C
- CKA
- Certified Kubernetes Administrator: a hands-on exam where you solve real tasks in a terminal against live clusters.
- Cluster
- The whole fleet: a set of machines (nodes) run together by one control plane.
- Explained at port 1.1 · The Cluster Map →
- ClusterIP
- The default Service type: a stable address reachable only from inside the cluster.
- Explained at port 1.16 · ClusterIP →
- Container
- A packaged app with everything it needs to run. Pods wrap containers.
- Explained at port 1.10 · Pods →
- Control plane
- The harbor office of the cluster: it decides what runs where and keeps the cluster in the state you asked for.
- Explained at port 1.1 · The Cluster Map →
- Controller
- A quartermaster: a loop that watches the cluster, compares it with what you asked for, and fixes the difference.
- Explained at port 1.6 · Controllers: the control loop →
- CRI
- Container Runtime Interface: the standard way the kubelet talks to a runtime such as containerd.
- Explained at port 1.2A · Did Kubernetes drop Docker? →
- crictl
- A command-line tool on a node that talks to the runtime through the CRI to list pods and containers.
- Explained at port 1.2B · ctr vs nerdctl vs crictl →
D
- Deployment
- Manages ReplicaSets for you, so you can roll out a new version of an app and roll it back.
- Explained at port 1.13 · Deployments →
- Dry run
- --dry-run=client -o yaml prints the YAML of an object without creating it.
- Explained at port 1.19 · Imperative vs declarative →
E
- etcd
- The cluster's logbook: a key-value store that holds the state of every object.
- Explained at port 1.3 · etcd: a key-value store →
K
- kube-apiserver
- The radio room: every request to the cluster goes through it, and only it reads and writes etcd.
- Explained at port 1.5 · kube-apiserver: what happens on kubectl create →
- kube-proxy
- Runs on every node and turns a Service's virtual IP into forwarding rules that reach the pods.
- Explained at port 1.9 · kube-proxy →
- kubectl
- The command-line tool you use to talk to a cluster: create, inspect, change and delete its objects.
- Explained at port 1.19 · Imperative vs declarative →
- kubectl apply
- Creates or updates objects from a file by comparing your file, the live object and the last-applied annotation.
- Explained at port 1.20 · kubectl explain & apply →
- kubelet
- The agent on every node that starts Pods and reports back to the control plane.
- Explained at port 1.8 · The kubelet →
L
- Label
- A key-value sticker on an object. Selectors use labels to pick the objects they manage.
- Explained at port 1.12 · ReplicaSets & labels →
- LoadBalancer
- A Service type that asks your cloud for a load balancer with one public IP, built on a NodePort.
- Explained at port 1.17 · LoadBalancer →
M
- Manifest
- A YAML file that describes an object: apiVersion, kind, metadata and spec.
- Explained at port 1.11 · A pod in YAML →
- Manual scheduling
- Placing a Pod on a node yourself by setting its nodeName, skipping the scheduler.
N
- Namespace
- A named section of the cluster that keeps teams or apps apart.
- Explained at port 1.18 · Namespaces →
- Node
- One machine in the cluster, a ship in the fleet. Your Pods run on nodes.
- Explained at port 1.1 · The Cluster Map →
- NodePort
- A Service type that opens the same port (30000-32767) on every node.
- Explained at port 1.15 · NodePort →
P
- Pod
- The smallest thing Kubernetes runs: one or more containers that share a network address and storage.
- Explained at port 1.10 · Pods →
- Practice cluster
- A small cluster on your own machine (for example kind or minikube) where you can break things safely.
R
- ReplicaSet
- Keeps a set number of identical Pods running, and replaces any that disappear.
- Explained at port 1.12 · ReplicaSets & labels →
- Rolling update
- Swapping old Pods for new ones a few at a time, so the app stays up during an upgrade.
- Explained at port 1.13 · Deployments →
S
- Scheduler
- The part of the control plane that picks a node for each new Pod.
- Explained at port 1.7 · The scheduler: filter & score →
- Service
- A steady address in front of a changing group of Pods, so others can always reach them.
- Explained at port 1.14 · Services: why pods need an address →
T
- Taint
- A mark on a node that keeps Pods away unless they tolerate it.
- Toleration
- A setting on a Pod that lets it land on a node with a matching taint.