World K5: Lunch Rush

Pick a stall: nodeSelector and affinity

You will learn: how to send a tray only to stalls with the right label.

Checked with Kubernetes 1.37 on 31 Oct 2026

The problem

You are here: cluster kind-dev namespace default

~/shop $ kubectl apply -f fast.yamlpod/fast created~/shop $ sleep 5~/shop $ kubectl get pod fastNAME   READY   STATUS    RESTARTS   AGEfast   0/1     Pending   0          5s

The fast tray says Pending. It never starts. What is it waiting for?

Watch

fast.yaml carries a sign. A says: “only stalls with this label.”

spec:
  nodeSelector:
    disk: ssd

Which stall may take the fast tray?

Try it

Put every tray on a stall. Read the signs first.

Tap a tray, then tap a stall. Or drag the tray.

Waiting trays

stall-a
disk=ssd
CPU 0 of 2
Memory 0Mi of 4Gi
stall-b
disk=hdd
CPU 0 of 2
Memory 0Mi of 4Gi
Pending
Trays that fit on no stall wait here.

Check

You are here: cluster kind-dev namespace default

~/shop $ kubectl apply -f fast.yamlpod/fast created~/shop $ sleep 5~/shop $ kubectl get events --field-selector reason=FailedScheduling -o jsonpath={.items[0].message}0/3 nodes are available: 1 node(s) had untolerated taint(s), 2 node(s) didn't match Pod's node affinity/selector. preemption: 0/3 nodes are available: 3 Preemption is not helpful for scheduling.

Those are the scheduler’s real words. In plain words: one stall is locked. Two have the wrong label. No stall says disk=ssd.

Try it

Give stall dev-worker2 the sign. The waiting tray should move in.

kubectl label nodes dev-worker2 disk=ssd

You are here: cluster kind-dev namespace default

dev-control-planeOpenManager

No trays here.

dev-workerOpen

No trays here.

dev-worker2Open

No trays here.

Waiting for a stall

fastPending

Time: +0s (sped up)

Watch

is a wish list: must-haves (required) and nice-to-haves (preferred). The fast tray needs ssd. It would like the front zone. Sort them.

Drag a card into a box. Or tap the card, then tap the box.

In YAML, the two lists look like this:

affinity:
  nodeAffinity:
    requiredDuringSchedulingIgnoredDuringExecution:
      nodeSelectorTerms:
        - matchExpressions:
            - { key: disk, operator: In, values: [ssd] }
    preferredDuringSchedulingIgnoredDuringExecution:
      - weight: 50
        preference:
          matchExpressions:
            - { key: zone, operator: In, values: [front] }

Predict

The fast tray runs on dev-worker2. Now you remove the label: kubectl label nodes dev-worker2 disk-. What happens to the tray?

You did it!

You can now send a tray to the right stall with labels.

Go deeper (optional)

kubectl get nodes --show-labels lists every stall’s labels. Built-in ones include kubernetes.io/hostname and, in clouds, topology.kubernetes.io/zone. Pod affinity puts trays near other trays. Anti-affinity keeps them apart.

Real rep (optional): use a throwaway kind cluster. Label a worker, then apply a pod with a nodeSelector.