Prepare and test your skills

Prepare and test your skills

Free sample question with answers and detailed explanation below. The Professional Cloud Network Engineer practice bank has an additional 2,437 practice questions.Try 10 free realistic questions: instant score, no credit card, and no account required.
Your organization manages a GKE Standard cluster running version 1.29 in a standalone VPC network. The cluster currently uses kube-dns with custom stubDomains configured in the kube-dns ConfigMap. You need to migrate to a fully managed DNS architecture that removes the operational overhead of dedicated DNS Pods while meeting the following requirements:
kube-dns without causing DNS resolution downtime for running workloads.kube-dns ConfigMap stubDomains configuration without manually recreating forwarding zones in Cloud DNS.kube-dns Pods remain running in the cluster after the migration completes.Which migration strategy should you implement?
kube-dns, configure an inbound Cloud DNS server policy for the VPC, and reduce the kube-dns-autoscaler min replica count to zero so NodeLocal DNSCache handles cluster and VPC queries locally.NodeLocal DNSCache is a GKE add-on deployed as a DaemonSet that runs a local caching DNS agent (169.254.20.10) on every node. When paired with kube-dns, it caches responses locally on the node to lower lookup latency, avoid conntrack table exhaustion, and reduce query load on central kube-dns Pods.
kube-dns is strictly limited to in-cluster resolution and cannot publish or resolve headless Service endpoint IP addresses across the VPC or to on-premises networks, even with an inbound Cloud DNS server policy.cluster.local still forward via TCP to the kube-dns-upstream Service backed by kube-dns Pods. Additionally, NodeLocal DNSCache itself runs as Pods on every node.kube-dns-autoscaler min replicas to zero (or scaling kube-dns to zero) prevents NodeLocal DNSCache from resolving uncached or expired cluster.local records.NXDOMAIN responses for 5 seconds.169.254.169.254) while sending cluster.local queries to kube-dns when Cloud DNS for GKE is not enabled.kube-dns Service.Enabling NodeLocal DNSCache alongside kube-dns would be appropriate if your sole goal was reducing internal Pod lookup latency and conntrack exhaustion inside a single cluster without requiring VPC-wide headless Service discovery or migrating away from in-cluster DNS Pods.
.local, upgrade all node pools to a newer GKE version, and then scale the kube-dns Deployment and kube-dns-autoscaler to zero.Cloud DNS VPC scope configures a GKE cluster so that its Service and Pod DNS records are published into a VPC-scoped private managed zone using a unique custom domain suffix (such as .cluster1) instead of cluster.local. This makes all cluster DNS names directly visible across the entire VPC network and to connected on-premises environments.
kube-dns to VPC scope is a disruptive operation. Because kube-dns cannot resolve queries targeting the newly assigned custom domain, Pods experience DNS resolution failures until every node pool has been re-created.stubDomains: VPC scope ignores the kube-dns ConfigMap completely. Any existing stubDomains or upstreamNameservers entries would stop functioning unless you manually recreate them as Cloud DNS forwarding zones.This approach would be appropriate if you were provisioning a brand-new GKE Standard cluster that required full VPC-wide resolution for all Services, you were willing to configure custom forwarding zones directly in Cloud DNS instead of relying on the kube-dns ConfigMap, or a maintenance window allowed disruptive DNS downtime during migration.
kube-dns Deployment and kube-dns-autoscaler to zero.When Cloud DNS is enabled on an existing GKE Standard cluster, the clouddns-controller begins syncing DNS records to Cloud DNS on the control plane, but existing nodes and Pods continue using the kube-dns Service IP in /etc/resolv.conf until the node pools are re-created with Cloud DNS configuration. Running gcloud container clusters upgrade while targeting the same GKE version already running on the node pool does not perform the required node re-creation for Cloud DNS adoption.
kube-dns Service IP (10.x.y.10) instead of 169.254.169.254.kube-dns: Immediately scaling the kube-dns Deployment and kube-dns-autoscaler to zero removes the backend Pods serving 10.x.y.10, causing immediate DNS resolution failures across the cluster.kube-dns running on GKE 1.19 and later specifically to prevent outages during node transitions.kube-dns to Cloud DNS, you must either upgrade the node pools to a new GKE version or create a new node pool (even on the same version) and migrate workloads./etc/resolv.conf inside Pods (nameserver 169.254.169.254, or 169.254.20.10 with NodeLocal DNSCache) confirms the transition is complete before disabling kube-dns.This approach would succeed only if the cluster created new node pools on the same GKE version (or upgraded existing node pools to a newer GKE version) and verified that all Pods updated their nameserver entry before scaling kube-dns down to zero.
.local, and then scale the kube-dns Deployment and kube-dns-autoscaler to zero.Cloud DNS for GKE replaces cluster-hosted kube-dns servers with a control-plane clouddns-controller that syncs Kubernetes Service and Pod records into managed private DNS zones and response policy zones. Pods send DNS queries to the local Compute Engine metadata server (169.254.169.254), eliminating dedicated in-cluster DNS Pods. Enabling additive VPC scope alongside cluster scope publishes headless Service records into an additional VPC-scoped private zone without altering internal cluster.local resolution.
kube-dns to cluster scope preserves the exact *.cluster.local resolution behavior, allowing seamless migration without application changes or downtime. Enabling additive VPC scope also occurs without impacting DNS uptime.stubDomains preservation: Unlike VPC scope, cluster scope automatically translates stubDomains and upstreamNameservers from the kube-dns ConfigMap into Cloud DNS forwarding zones (applying to both Pods and nodes)..local, which causes resolution failures) so VPC VMs and on-premises clients over Cloud Interconnect can resolve Pod endpoint IPs directly.kube-dns Pods: Upgrading node pools to a new GKE version updates Pod /etc/resolv.conf files to point to 169.254.169.254. Once no workloads rely on kube-dns, scaling the kube-dns Deployment and kube-dns-autoscaler to zero safely removes all DNS Pods.kube-dns Pods.169.254.169.254.This sequence avoids the disruptive downtime and ConfigMap incompatibility of VPC scope, ensures nodes actually transition off kube-dns through a version upgrade before Pod removal, and achieves hybrid headless Service discovery.
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