A machine in a cluster, physical or virtual, the unit of capacity Kubernetes and similar orchestrators schedule onto.
A node is a single machine, physical or virtual, that participates in a cluster. In Kubernetes, a node runs a kubelet that registers with the control plane, accepts pods, and reports back resource usage. Nodes can be drained (gracefully evicted) for maintenance, cordoned (no new pods scheduled) for diagnostic, or terminated (aggressively removed) by autoscaling. The node lifecycle (provision, join, drain, terminate) is the fundamental capacity primitive every cluster operator works with.
Most cluster incidents trace back to node-lifecycle events: a node ran out of disk and stopped scheduling, autoscaling terminated a node that still had pods on it, a node-image update broke the kubelet. Treating nodes as cattle (replaceable, identical, owned by automation) instead of pets (named, hand-tuned, irreplaceable) is the operational discipline that makes Kubernetes scale.
See the part of the platform that handles node (cluster node) in production.