He pushed it to the team’s dotfiles repo with a single commit message: From that day on, every engineer on the team had a colorful, impossible-to-ignore shell prompt. And not once—not a single time—did anyone ever again spend ten minutes troubleshooting the wrong cluster.
kubectl logs payment-api-f9k3l-8hj2s -n payment-system --previous There it was. A connection pool exhaustion error to the Redis cache. The new deployment had increased the default pool size, and the production Redis max connections was too low.
Now he ran the same commands.
He switched contexts immediately:
He closed his laptop, leaned back, and stared at the ceiling. kubectl switch contexts
CURRENT NAME CLUSTER NAMESPACE prod-us-east prod-cluster payment-system * staging-us-west staging-cluster payment-system dev-local dev-cluster default He was looking at . The alert was from production .
He tried hitting the service endpoint manually, using port-forward: He pushed it to the team’s dotfiles repo
kubectl config get-contexts The output made his stomach drop: