liveBook · Manning

Pod ip address

As explained in Google Anthos in Action

A pod IP address is the network address assigned to a Kubernetes pod for communication within its cluster.

Every Pod in an Anthos GKE cluster receives a Pod IP address. Pod IPs are routable within the cluster; whether they are reachable from outside depends on the cluster implementation. They use an address range separate from those for nodes and API servers, and for Services. section 10.2.1

The API server divides a Pod CIDR into unique ranges for nodes. A Pod receives an address from its node’s range, so its address changes if it moves to another node. Size the Pod CIDR for the maximum node count and the planned Pods per node. section 10.2.1

Plan for 30 Pods per node

Each node needs at least a /27 range, containing 32 addresses. The cluster Pod range must accommodate one such range for every node at the planned maximum size. section 10.2.1

Do not rely on a Pod IP as a stable Service address. A Service’s ClusterIP remains static and acts as a virtual IP for multiple Pods, even though individual Pod IPs can change. section 10.2.1

How Google Anthos in Action covers it

How deep it goes
Explains per-node CIDR allocation and capacity planning, while deferring details of external routability.
Examples
Uses 30 Pods per node to illustrate a minimum /27 per-node range.
What it assumes
Assumes familiarity with Pods, nodes, and CIDR notation; points elsewhere for IP subnet basics.
What you can do after
Operate/configure it by sizing a Pod CIDR for planned Pods per node and maximum node count.
How it connects
Connects Pod CIDRs to node IP ranges, API servers, Kubernetes Services, and ClusterIP addresses.

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.