Labels, Selectors and nodeSelector

Scheduling and Placement / Guide 39 of 103 / Beginner

• Kubernetes1.36.4
• Cluster4 nodes
• Runtimecontainerd 2.2.6
• CNICalico v3.32.1
• TimeAbout 25 min

Node names and ages are this lab's. The built-in label set is standard.

This guide includes

Use this when one object has to find another - a Service finding Pods, a Deployment tracking what it owns, a Pod landing on the right node. This matters because labels are the only addressing system Kubernetes has: nothing here finds anything by name.

• reading the labels the cluster has already put on its own nodes
• adding labels of your own, which count for exactly as much as the built-in ones
• selecting four different ways, since -l takes far more than equality
• placing a workload with nodeSelector, the simplest tool that uses them
• removing a label with a trailing hyphen, the same convention taints use

Before you start

• A cluster with a few nodes and kubectl.
• Permission to label nodes, which is cluster-scoped.
• Nothing else. This is the mechanism everything else in Kubernetes is built on, so it is worth doing early.

1. 1

   Your cluster is already covered in labels

2. 2

   Add your own

3. 3

   Four ways to select

4. 4

   Use a label to place a workload

5. 5

   Remove the labels

Official sources

• Labels and selectors
• Well-known labels, annotations and taints
• Assigning pods to nodes
• Recommended labels