He stared at the screen. One wrong kubectl apply and he could charge a cactus to someone’s credit card.
kubectl config use-context prod-dallas But instead of applying immediately, he paused. He wrote a tiny bash alias:
It was Vlad’s first day on the job as a platform engineer at a chaotic startup called Nebulous Systems. The previous “kube-whisperer” had left behind a labyrinth of Kubernetes clusters: staging, prod, legacy, and one ominously named “do-not-touch.”
CURRENT NAME CLUSTER AUTHINFO minikube minikube minikube * prod-dallas prod-cluster-dallas admin-prod staging-eu staging-cluster-eu deployer legacy-bare-metal legacy-cluster old-admin The asterisk told him he was currently pointing at . His heart rate doubled. That was production . Live traffic. Real money. Customers ordering cat-shaped planters at 2 AM.
kubectl config get-contexts The output appeared like a cryptic map:
His onboarding task was simple: “Roll out the fix for the payment service.”
That night, Vlad added a sticky note to his monitor: And from that day on, not once did he deploy staging YAML to prod—or worse, to the do-not-touch cluster that, rumor had it, still ran the original prototype of the cat planter checkout flow from 2018.
