You run a Deployment with 5 replicas for a critical API on AKS. You must ensure that at least 4 replicas remain available during planned disruptions such as node OS upgrades and during rolling updates of the Deployment. What should you configure?

The correct option ensures high availability during planned disruptions and rolling updates. A PodDisruptionBudget (PDB) with minAvailable: 4 guarantees that at least four replicas are running, even when Kubernetes voluntarily evicts pods (e.g., during node maintenance). The Deployment's rollingUpdate strategy with maxUnavailable: 1 means only one pod can be unavailable at any given time during an update, and maxSurge: 1 allows one extra pod to be created before an old one is terminated, speeding up the update while maintaining capacity. The second option is incorrect because maxUnavailable: 2 in the PDB would allow too many pods to be unavailable, violating the requirement of at least 4 available. Setting maxSurge: 0 would slow down updates. The third option is incorrect because not using a PDB offers no protection against voluntary evictions. Setting maxUnavailable: 0 would prevent any pod from being unavailable, which is often too restrictive for rolling updates and can lead to deadlocks if not enough resources are available for maxSurge. The fourth option, Recreate strategy, is unsuitable for high availability as it terminates all old pods before creating new ones, leading to downtime.