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 connect vpc-compute (MTU 8896) and vpc-services (MTU 1460) via VPC Network Peering. A multi-NIC Linux VM (app-node-1) has nic0 in vpc-compute (8896 MTU) and nic1 in vpc-mgmt (1460 MTU), with firewall rules blocking unsolicited ingress ICMP.
Sending 8,000-byte UDP packets (DF bit set) from nic0 to an internal passthrough Network Load Balancer (UDP forwarding rule) in vpc-services causes silent drops:
app-node-1 must keep using 8,896-byte frames inside vpc-compute.vpc-services must dynamically adapt its packet size.Which actions should you take to resolve this issue?
vpc-compute permitting ICMP from the destination IP range in vpc-services, and recreate the internal passthrough Network Load Balancer forwarding rule in vpc-services using the L3_DEFAULT protocol.L3_DEFAULT forwarding rules doEnabling Path MTU Discovery (PMTUD) across peered VPC networks with mismatched MTUs requires both an ingress ICMP allow rule on the sending VPC and an L3_DEFAULT forwarding rule on the receiving internal passthrough Network Load Balancer. When app-node-1 transmits an 8,000-byte UDP packet with the Don't Fragment (DF) bit set toward vpc-services, Google Cloud drops the oversized packet because it exceeds the 1460-byte MTU of vpc-services and generates an ICMP Type 3, Code 4 (Fragmentation Needed) message (or ICMPv6 Packet Too Big for IPv6) back to the sender.
vpc-compute: Keeping nic0 at 8896 bytes allows app-node-1 to continue high-throughput jumbo frame communication with peer instances inside vpc-compute while maintaining proper multi-NIC MTU alignment (nic0 at 8896, nic1 at 1460).L3_DEFAULT allows the load balancer to process both the original UDP datagrams and the ICMP control traffic associated with the VIP. Allowing ingress ICMP in vpc-compute from the destination IP range in vpc-services permits the Fragmentation Needed message (whose source IP matches the original destination VIP) to reach app-node-1's kernel, which lowers the effective path MTU for that destination.1460-byte path MTU specifically for the vpc-services VIP without degrading intra-VPC 8896-byte flows.L3_DEFAULT ensures ICMP error signaling works seamlessly alongside UDP traffic on passthrough forwarding rules.vpc-services maintains strict security boundaries.vpc-compute nor app-node-1 needs to be stopped or reconfigured with a lower interface MTU.Unlike lowering nic0's MTU or altering the VPC MTU—which breaks intra-VPC jumbo frame performance—combining ingress ICMP allowance with an L3_DEFAULT forwarding rule directly fixes the broken PMTUD feedback loop while preserving each NIC's native VPC MTU.
nic0 interface MTU on app-node-1 to 1460 bytes inside the Linux OS to match nic1, and add an egress allow firewall rule in vpc-compute permitting ICMP toward vpc-services.nic0 MTU and adding egress ICMP doesManually setting nic0 to 1460 bytes inside the Linux guest OS forces the network interface to cap all outbound IP packets at 1460 bytes regardless of the attached VPC MTU (8896). Meanwhile, adding an egress allow firewall rule for ICMP in vpc-compute permits outbound ICMP packets from app-node-1 toward vpc-services.
nic0's MTU to 1460 bytes caps all traffic leaving nic0, including flows destined for other VMs inside vpc-compute. This prevents app-node-1 from exchanging 8,896-byte jumbo frames within vpc-compute.vpc-compute only governs outbound ICMP packets generated by app-node-1. It does not allow inbound (ingress) ICMP Type 3, Code 4 (Fragmentation Needed) messages returning from vpc-services.L3_DEFAULT forwarding rule requirement: Even if PMTUD were used, a UDP-only internal passthrough Network Load Balancer forwarding rule cannot process ICMP PMTUD traffic.8896 for nic0 and 1460 for nic1.Fragmentation Needed packets are generated in response to dropped UDP datagrams (not established ICMP sessions), an ingress allow rule for ICMP is required on the sender's VPC.app-node-1 did not require 8,896-byte jumbo frames inside vpc-compute, and the goal were solely to cap all nic0 traffic at 1460 bytes—though an egress ICMP rule would still not enable PMTUD.
While setting a NIC MTU lower than its VPC MTU is technically supported for specialized scenarios, it sacrifices intra-VPC jumbo frame throughput when applied globally to an interface.
vpc-services permitting ICMP from vpc-compute, and configure the sending application on app-node-1 to clear the Don't Fragment (DF) bit on outbound UDP packets.This approach attempts to resolve MTU-related drops by permitting ingress ICMP on the receiving VPC (vpc-services) and disabling the Don't Fragment (DF) bit on outbound UDP datagrams generated by app-node-1. The underlying assumption is that clearing the DF bit allows intermediate network hops to fragment oversized packets, and that ICMP rules belong on the destination network.
Fragmentation Needed) messages travel back to the sender (app-node-1) with a source IP matching the original destination VIP. An ingress rule in vpc-services does not allow those error packets to enter vpc-compute, so app-node-1 remains blind to the MTU bottleneck.DF bit does not trigger in-transit IP fragmentation, and Google Cloud still drops packets exceeding 1,600 bytes on a VPC with an MTU below 1,600 bytes.UDP prevents it from processing ICMP packets required for PMTUD signaling.1,600 bytes and an incoming packet exceeds 1,600 bytes, Google Cloud drops it and sends Fragmentation Needed even when the DF bit is off.vpc-compute).The ingress ICMP allow rule were placed in vpc-compute (from vpc-services sources), the forwarding rule used L3_DEFAULT, and the application explicitly sized UDP payloads to 1,432 bytes or relied on PMTUD rather than clearing the DF bit.
A frequent networking misconception is placing ICMP rules on the destination VPC rather than the sender's VPC, and expecting intermediate routers to fragment packets with DF=0.
vpc-compute to automatically fragment oversized UDP packets at the peering boundary.TCP Maximum Segment Size (MSS) clamping is a mechanism where an intermediate network device rewrites the MSS option inside TCP SYN and SYN-ACK packets so segments fit within a downstream link's MTU. IP fragmentation splits an oversized IP packet into smaller packets when traversing a link with a smaller MTU.
app-node-1 is transmitting UDP datagrams, which are connectionless and have no SYN/SYN-ACK handshake or MSS header field.Fragmentation Needed messages.No configuration in Google Cloud enables UDP fragmentation or MSS clamping on VPC Network Peering; instead, hybrid paths like Cloud VPN or Cloud Interconnect VLAN attachments use MSS clamping for TCP and pre-encapsulation PMTUD signaling for non-TCP traffic.
A common misconception is treating VPC Network Peering like a traditional Layer 3 WAN router or VPN gateway that performs MSS clamping and IP fragmentation across network boundaries.
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