Friday, March 15th, 2024
Persuasion is often more effectual than force.
— Aesop
- Kubernetes crash course
- Kubernetes distributes containerized services / workloads across a cluster (set of nodes) & manage lifecycle of containers given a declarative configuration of desired state
- node: a physical machine or VM
- cluster: a collection of nodes
- control plane: set of master nodes making up the central components & processes of a Kubernetes cluster (e.g., API server, scheduler, controller manager, etcd)
- master node: a node that runs key K8s processes; it is recommended to have at least a few replicas to ensure HA
- worker node: a node that runs K8s pods; worker nodes are usually better spec’d than master nodes due to its high workload
- pod: an abstraction over containers (only Docker images are supported for now)
- Although pods can contain multiple containers, usually a pod only holds one container.
- Each pod can get their own IP in the virtual network managed by K8s
- pods can be removed and recreated, e.g., when they are unhealthy
- service: gives a group of (replica) pods the same IP
- ???
- deployments
- ???
- replicaset
- statefulset
- ???set ???