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.
You manage a production VPC-native Google Kubernetes Engine (GKE) cluster in us-central1. Network Analyzer flags a high IP utilization insight (google.networkanalyzer.container.ipAddressInsight) reporting 87.5% Pod IP allocation, even though only 550 Pods are running.
Your current network configuration includes:
/24): 15% utilized/21): 2,048 IPs total, 7 active nodes--max-pods-per-node=110, with each node running at most 25 PodsYou must scale out by 15 additional nodes without recreating the cluster or causing workload downtime. Sufficient RFC 1918 space remains in the VPC.
Which strategy should you implement?
Cloud NAT dynamic port allocation optimizes how a Cloud NAT gateway assigns external NAT IP addresses and source port tuples to VM instances and GKE nodes for outbound internet traffic. Lowering the minimum ports per VM and disabling endpoint-independent mapping reduces wasted public NAT ports on idle or low-traffic nodes.
nat_allocation_failed). It has zero effect on internal VPC secondary Pod alias IP ranges or how GKE carves out RFC 1918 Pod CIDR blocks per node./21 secondary range. The cluster autoscaler will still fail to schedule more than one additional node once the remaining 256 Pod IPs are consumed.compute.googleapis.com/nat/port_usage) resolves egress SNAT port exhaustion and reduces the number of external static or auto-allocated NAT IP addresses required.The utilization warning or failure involved nat_allocation_failed alerts from Cloud NAT because nodes scaling out exhausted available external NAT IP addresses and source ports while connecting to public internet endpoints.
Engineers troubleshooting scaling bottlenecks must distinguish between egress NAT capacity warnings (solved via dynamic port allocation or adding external NAT IPs) and internal VPC alias IP exhaustion flagged by Network Analyzer's ipAddressInsight (solved via multi-Pod CIDR and node pool right-sizing).
gcloud compute networks subnets expand-ip-range to expand the subnet's secondary Pod IPv4 range from /21 to /19 in place so existing and new nodes use the expanded CIDR.expand-ip-range doesThe gcloud compute networks subnets expand-ip-range command increases the size of a VPC subnet's primary IPv4 host address range by shortening its prefix length (for example, expanding a /24 to a /22) without disrupting running VM instances or recreating resources.
expand-ip-range. Attempting to modify the secondary Pod /21 range directly in place fails./24, 252 usable host IPs) is only at 15% utilization and already has plenty of capacity for 22 total nodes. The exhaustion bottleneck is strictly in the secondary Pod CIDR range, where 7 nodes with --max-pods-per-node=110 consume 1,792 of the 2,048 available IPs.expand-ip-range applies exclusively to a subnet's primary IPv4 CIDR block and is irreversible once applied.The primary IPv4 range of the subnet were approaching exhaustion due to too many Compute Engine VMs, internal load balancer forwarding rules, or GKE node primary interfaces, rather than the secondary Pod alias IP range.
A common misconception is that VPC subnet expansion commands apply uniformly to both primary and secondary ranges. In Google Cloud, only primary ranges grow in place; secondary Pod IP capacity must be scaled horizontally by attaching additional secondary ranges with discontiguous multi-Pod CIDR.
--max-pods-per-node from 110 to 32 so GKE dynamically reclaims unused Pod IP addresses from each running node's /24 CIDR block.--max-pods-per-node configuration doesThe --max-pods-per-node setting controls the size of the contiguous Pod CIDR block that GKE reserves from a subnet's secondary IPv4 range whenever a new node is provisioned in a node pool.
/24 CIDR block in place. Even though only 550 of 2,048 Pod IPs are actively bound to Pods, all 1,792 IPs across the 7 /24 blocks remain exclusively locked to those nodes.--max-pods-per-node on an existing node pool to retroactively resize Pod CIDRs. Only 1 /24 block (256 IPs) remains unallocated in the /21 range, which cannot accommodate 15 additional nodes.2^(ceil(log2(2 * max_pods_per_node))) addresses per node to prevent rapid IP recycling collisions during Pod rollouts.--max-pods-per-node value and migrating workloads to it.You were provisioning a brand-new node pool on a cluster that still had enough unallocated secondary range capacity for all replacement nodes before draining the old pool—though even then, adding a new secondary range is safer when the existing /21 is already at 87.5% utilization.
Understanding the difference between allocated node CIDR blocks and in-use Pod IPs is critical when interpreting Network Analyzer utilization warnings, as high allocation ratios stall cluster autoscalers long before actual Pod counts approach the subnet total.
--additional-pod-ipv4-ranges, create a new node pool configured with that range and --max-pods-per-node=32, and cordon and drain workloads from the old node pool before deleting it.Discontiguous multi-Pod CIDR allows you to attach additional secondary IPv4 ranges to an existing VPC-native GKE cluster without recreating the cluster or provisioning new subnets. Pairing this capability with a blue-green node pool migration lets you transition workloads cleanly onto right-sized node pools.
--max-pods-per-node=110, GKE allocates 2^(ceil(log2(2 * 110))) = 256 IPs (/24) per node, so 7 nodes consume 1,792 of 2,048 IPs (87.5%), leaving room for only 1 more node. Lowering --max-pods-per-node to 32 reduces each node's Pod CIDR allocation to 64 IPs (/26)—a 4x reduction in IP consumption while still exceeding the 25-Pod per-node requirement.--additional-pod-ipv4-ranges immediately unlocks ample Pod IP space for the new node pool.--pod-ipv4-range), gracefully cordoning and draining old nodes, and deleting the old pool avoids cluster rebuilds and keeps production workloads available throughout the migration./21 secondary range for future reuse.google.networkanalyzer.container.ipAddressInsight alert once migration completes.Unlike in-place subnet expansion (which cannot resize secondary ranges) or mutating existing node pools (which cannot shrink already-assigned node CIDRs), combining --additional-pod-ipv4-ranges with a blue-green node pool migration solves both immediate capacity limits and structural IP over-allocation without downtime.
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