s0sta
EN NL عربي
Filters
How to search
  • Plain words find names, texts and examples.
  • "two words" finds the exact phrase.
  • --name or -x finds an option.
  • #word looks only inside commands.
  • @grep finds lessons and labs that use grep.
  • Filters: cat:network · in:labs · in:notes · level:rhcsa · ch:8 · root:yes. Put ! before a word to leave it out. OR between words means one of them.
  • ↑ ↓ to move · Enter to open · Esc to close · / to search

KubernetesPythonFixes series

Quiz: 3 Kubernetes failure questions

Three questions separate a kubectl guess from a diagnosis, and all three come back in a review.

Quick reference

Q1. Why does a Pending pod have no logs?

  • a) It never started a container
  • b) The log driver is disabled
  • c) The pod is on an unscheduled node

Answer: a) It never started a container. A Pending pod has not been placed on a node, so no image was pulled and no process ran. The evidence is the FailedScheduling event, not a log.

Q2. What does an eviction reason tell you?

  • a) Which node resource ran out
  • b) Which container crashed
  • c) Which probe failed

Answer: a) Which node resource ran out. Evicted means the kubelet killed the pod to protect the node. The message names memory, storage or pid, which points at capacity or a leak, not at the app.

Q3. Why does CrashLoop backoff grow?

  • a) To give the node time to recover
  • b) To slow a failing restart loop
  • c) To wait for the image pull

Answer: b) To slow a failing restart loop. Each restart doubles the wait from 10 seconds up to 5 minutes. It protects the kubelet and the network, and a fast fix still waits for the current backoff.

Explained

Question one

Why does a Pending pod have no logs? Because it never started a container. The scheduler could not place it, so no image was pulled and no process ran. The FailedScheduling event is the evidence; a log search on a Pending pod returns nothing, and that nothing is expected.

Question two

What does an eviction reason tell you? Which node resource ran out. Evicted means the kubelet killed the pod to protect the node, and the message names memory, ephemeral-storage or pid. It is a node-level event, not an application error, so the pod's exit code is not the place to look. The node, not the app, is the first suspect.

Question three

Why does CrashLoop backoff grow? To slow a failing restart loop. Each restart doubles the wait from 10 seconds up to 5 minutes. It protects the kubelet and the network, and a fast fix still waits for the current backoff. The delay resets only after a stable run, not after a fix.

The honest caveat: backoff and eviction are both protection mechanisms, one for the restart loop and one for the node. Neither fixes the underlying fault, so the fault is still yours to find.

The lesson: know the phase, skip the guess.

In one line

Know the phase, skip the guess. Three answers that save a wrong kubectl command.

Read the phase

kubectl get pod -n prod api -o wide

More notes: Kubernetes

Open in the notes app: search, filters, copy This note on GitHub