Prepare and test your skills
Prepare and test your skills
Free sample question, answered and explained below. The Professional Cloud DevOps Engineer practice bank has 1,455 more like it. Try 10 free: instant score, no account.
A DevOps team manages a fleet of GKE Standard clusters hosting mission-critical stateful workloads with long-lived active client connections. Following a blameless post-mortem regarding an outage during automated node patching and upgrades, the team discovered that GKE terminated pods when PodDisruptionBudget (PDB) constraints exceeded the default 1-hour timeout, and the team had no automated mechanism to validate workload stability before old nodes were deleted.
Which node pool upgrade configuration should the team implement to prevent forced pod terminations and enable rollback capabilities if anomalies occur?
A blue-green upgrade strategy in Google Kubernetes Engine (GKE) provisions an entirely new set of nodes (green pool) running the target Kubernetes version alongside the existing pool (blue pool). It controls workload migration incrementally and allows administrators to define validation criteria before decommissioning the old nodes.
Unlike surge upgrades—which mutate the existing pool and enforce rigid 1-hour PDB timeouts—blue-green upgrades decouple node provisioning from workload draining and give engineers full control over validation and rollbacks.
cluster-autoscaler.kubernetes.io/safe-to-evict: false.The Cluster Autoscaler location policy determines how new nodes are allocated across zones, while the cluster-autoscaler.kubernetes.io/safe-to-evict: false annotation prevents the Cluster Autoscaler from terminating a node during normal scale-down operations if that specific pod is running on it.
safe-to-evict annotation only guards against voluntary scale-down triggered by the Cluster Autoscaler when nodes are underutilized. It does not govern GKE node pool upgrade processes, node auto-repair, or control-plane maintenance.safe-to-evict annotation, enforcing the standard 1-hour PDB timeout.The goal was to prevent the Cluster Autoscaler from consolidating or deleting underutilized nodes running non-replicated, restart-intolerant batch processing pods during daily operations.
maxSurge=1 and maxUnavailable=0, and configure a PodDisruptionBudget with minAvailable=100%.Surge upgrades update nodes in-place by adding a configurable number of extra surge nodes (maxSurge) and replacing existing nodes while respecting maxUnavailable. A PodDisruptionBudget (PDB) with minAvailable=100% instructs the Kubernetes eviction API to prevent voluntary disruptions that would drop total replica count below 100%.
minAvailable=100% or maxUnavailable=0 is specified, GKE forcefully terminates and deletes remaining pods after 60 minutes to prevent the upgrade from hanging indefinitely.minAvailable=100%) risk stalling normal node drain sequences until the timeout forces pod termination.The workload consisted purely of stateless services that could quickly reschedule to surge nodes within minutes and did not require extended soak periods or dedicated rollback pools.
gcloud container clusters resize and drain nodes during maintenance exclusions.This approach uses custom automation outside of GKE's native lifecycle manager to manually resize node pools up, trigger kubectl drain, and resize the old node pool down within scheduled maintenance exclusions.
You were managing an unmanaged vanilla Kubernetes cluster on raw Compute Engine VMs where native cloud platform upgrade controllers were unavailable.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated Validated