Lab: Pending Pods (Scheduling Constraints)

Goal

You will learn to diagnose Pending pods by reading events and the pod spec, then applying a smallest-safe fix (reduce requests, adjust constraints, or add capacity).

• Read scheduler events.
• Identify the blocking constraint.
• Change one governing input and verify placement.

Scenario

A pod is created and stays Pending. The team keeps re-applying the manifest. Nothing changes.

Your job is to prove why it cannot be scheduled.

Read the scheduler’s message

The scheduler emits a precise diagnosis. Find it and interpret it.

• Look for: Insufficient CPU/memory, taints, node selector mismatch, preemption notes.

kubectl describe pod <pod> -n <ns>
kubectl get events -n <ns> --sort-by=.lastTimestamp | tail -n 40

Inspect the governing constraints

Placement is governed by requests and constraints.

• Confirm requests are honest.
• Confirm selectors/taints/affinity match real nodes.

kubectl get pod <pod> -n <ns> -o yaml | rg -n "resources:|requests:|limits:|nodeSelector:|tolerations:|affinity:|topologySpreadConstraints"

kubectl get nodes -o wide

Resolution patterns

Prefer the fix that matches your intent and preserves safety.

• If requests were padded, reduce them and document why.
• If taints/selectors are wrong, correct them; avoid impossible affinity.
• If the cluster is truly out of capacity, add nodes or enable autoscaling with guardrails.