Full lesson
Explore the full explanation, examples, and visuals at your own pace.
A Pod is Kubernetes’ smallest schedulable unit: one or more containers.
A Pod is Kubernetes’ schedulable unit: it places one or more containers together on a node.
One application, two cooperating containers
The Web app and Log helper are separate containers inside one Pod. They’re grouped because the helper supports this particular application; it isn’t a shared logging service for every Pod.
Both containers use the Pod network
A Pod gives cooperating containers one network identity and one scheduling home. The web app and log helper share the Pod’s IP, so the app reaches the helper through localhost, using a different port.
The app sends logs to its helper
The client sends an HTTP request to the web app through the Pod’s network identity. Inside the Pod, the app sends logs to the helper over localhost, using the helper’s port, then returns an HTTP response. The client doesn’t use localhost to reach the helper.
How should the web app reach its logging helper when both run in the same Pod?
Let's think this through. How should the web app reach its logging helper when both run in the same Pod? A: Through localhost and the helper's port. B: Through a separate IP assigned to the helper. C: Through the Kubernetes node's IP. Choose an answer, or just think it through. I'll explain in a moment.
- Through localhost and the helper's port
- Through a separate IP assigned to the helper
- Through the Kubernetes node's IP
How should the web app reach its logging helper when both run in the same Pod?
The answer is A: Through localhost and the helper's port. Containers in one Pod share a network namespace, so localhost reaches the same Pod network. The helper listens on its own port; it does not get a separate Pod IP.
- Through localhost and the helper's port
- Through a separate IP assigned to the helper
- Through the Kubernetes node's IP
Shared network does not mean shared memory
The Pod contains the web app and log helper. They can both mount the same volume when it’s configured, but their memory stays separate. CPU and memory requests and limits are still set per container.
Kubernetes places the whole Pod on one node
Kubernetes schedules the Pod as one unit, not the web app and log helper separately. It checks whether a node can meet their combined resource requests, then keeps them together, so the app can reach its helper over localhost.
One network identity; separate containers
A Pod groups tightly coupled containers into one schedulable, networked unit. Here, the app reaches its logging helper through localhost, while each container keeps its own memory.