liveBook · Manning

Endpoint slice

As explained in Google Anthos in Action

An EndpointSlice is a Kubernetes resource that contains a subset of network endpoints for a Service.

An EndpointSlice is a Kubernetes resource that maps a Service to the IP addresses of matching application Pods. Kubernetes builds the mapping from the Service selector and matching Pod labels. section 3.2.6

Operational behavior:

  • A Service selector such as app=nginx-frontend identifies matching Pods in the namespace. section 3.2.6
  • Matching Pod IP addresses are added to EndpointSlices, while the Service provides a stable access point. section 3.2.6
  • When a Pod restarts and its IP changes, kube-controller-manager updates the endpoints. section 3.2.6

NGINX frontend

A Service selecting app=nginx-frontend discovers two matching Pods and adds their IP addresses to EndpointSlices. A related deployment scenario shows three Pod IPs, including 192.10.1.105, 192.10.3.107, and 192.10.4.108. section 3.2.6

How Google Anthos in Action covers it

How deep it goes
Explains selector matching, IP population, and controller updates, but does not detail EndpointSlice fields or scaling internals.
Examples
Uses an nginx-frontend selector, matching Pods, concrete Pod IP addresses, and a Kubernetes endpoints diagram.
What it assumes
Assumes familiarity with Kubernetes Services, Pods, labels, selectors, namespaces, and Pod IP addresses.
What you can do after
Diagnose/debug with it by tracing Service selectors to matching Pods and checking endpoint updates after Pod restarts.
How it connects
Connects Services, selectors, labels, Pods, Pod IP addresses, namespaces, and kube-controller-manager.
Cover of Google Anthos in Action
liveBookmanning.com
Read Google Anthos in Action on liveBook Get the book at manning.com

Also explained in

Compare how 2 books explain endpoint slice →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Google Anthos in Action. The section links point to the parts of the book it draws on.