Found this clear explainer on LinkedIn from DevOps Universe, and it's worth keeping handy: Kubernetes has three probes, and they each answer a different question.
Image: DevOps Universe, via the original LinkedIn post.
Startup probe — Has the app finished starting? It gives slow-starting apps time to boot, and liveness and readiness stay paused until it passes. If it never passes, Kubernetes restarts the container.
Liveness probe — Is the app still alive? It catches a stuck or deadlocked app, and on failure Kubernetes restarts the container.
Readiness probe — Is the pod ready to serve traffic? On failure the pod is removed from the Service so it stops receiving traffic, but it is not restarted.
Common mistakes the post calls out: using a liveness probe to check the database (one DB blip then restarts every pod), running slow apps with no startup probe (liveness kills them before they finish booting), and running with no readiness probe (users hit pods that aren't ready yet).
As the author puts it: "Startup protects the boot. Liveness protects against hangs. Readiness protects your users."
Read the original post on LinkedIn here: https://www.linkedin.com/posts/kubernetes-devops-k8s-share-7519379164957892609-O71O/
Credit: DevOps Universe on LinkedIn. This blog post is a summary with attribution, not a republication of the original.
