A Pod is the smallest thing Kubernetes runs

Study notes

• A container on its own is a process with an isolated view of the filesystem, network and process table. Kubernetes adds one level above it: the Pod. All containers in a Pod share a network namespace (one IP address, one port space, so they reach each other on localhost) and can mount the same volumes. They are always scheduled together to the same node, started and stopped as a unit.
• Why does this level exist? Because many real systems need a helper process tightly attached to the main one. A log shipper reads files the app writes, a service-mesh proxy intercepts the app's traffic, a config reloader watches a mounted file. Putting them in one Pod gives them exactly the shared resources they need without any special networking. This is the mechanical foundation of the sidecar pattern.
• The second idea is that Pods are ephemeral. You almost never create a bare Pod: a controller such as a Deployment creates Pods for you, and when one fails it is replaced, not repaired. The replacement has a new name and a new IP, which is why Services (Module 6) exist to give a stable address.
• Press the right arrow to reveal the structure: the app container, a sidecar, then the two things they share. Production gotcha: do not put two independent services into one Pod just to save a manifest. They would scale together, restart together and fail together.

Deck map