Pass the Linux Foundation Cloud & Containers Cilium-Associate Questions and answers with Dumpstech

Technical explanation

B expresses the principal scalability distinction intended by the question. In kube-proxy’s iptables mode, Kubernetes Services and endpoints are represented by chains of netfilter rules. As the number of Services and backends grows, rule creation, synchronization, and packet traversal can involve increasingly large rule sets. Matching through those chains is generally described as having linear scaling characteristics.

Cilium’s kube-proxy replacement stores service and backend state in eBPF maps. Hash-map lookups allow the datapath to locate service entries without sequentially scanning a rule for every Service, providing effectively constant-time lookup behavior for the relevant map operations. This makes the approach better suited to large and frequently changing Kubernetes environments.

Option A is false because iptables and netfilter also operate within the Linux kernel. Option C is false because kube-proxy’s iptables implementation supports IPv6 when the surrounding Kubernetes and host configuration supports it. Option D does not explain kube-proxy’s primary scaling limitation and introduces an unrelated SmartNIC/XDP claim.

The constant-time description is a useful architectural simplification; total performance still depends on map sizing, backend selection, connection tracking, and kernel behavior.

Official references

Cilium eBPF Datapath , Cilium Tuning Guide

Study Guide topic: kube-proxy replacement, eBPF maps, and datapath scalability.

Technical explanation

A is clearly the intended answer because its filters correspond to the displayed source namespace ( endor ), destination port ( 443 ), and verdict ( DROPPED ). However, it contains example.con , whereas the observed flow names example.com . Therefore, none of the options would literally return this exact flow if the FQDN filter is matched as written. The corrected command is:

hubble observe --to-fqdn example.com --from-namespace endor --to-port 443 --verdict DROPPED

Option B is malformed in two additional places and filters port 80 rather than 443. Option C selects the wrong FQDN and source namespace. Option D requests forwarded flows, directly contradicting the Policy denied DROPPED verdict; its wildcard also does not repair the verdict mismatch.

Hubble’s observe filters are cumulative: a returned flow must satisfy the specified destination FQDN, originating namespace, destination port, and verdict. Layer 7 visibility is enabled with a Cilium network policy containing the relevant L7 rules, which redirects selected traffic through the proxy so that application-level details can be reported.

Official references

Inspecting Network Flows with the Hubble CLI , Layer 7 Protocol Visibility

Study Guide topic: Hubble flow filtering, FQDN visibility, namespaces, ports, and verdicts.

Technical explanation

In cluster-scope IPAM, the Cilium Operator allocates a PodCIDR to each node from the configured cluster-wide address pool. It records those allocations in each node’s CiliumNode custom resource, specifically under spec.ipam.podCIDRs . The Cilium agent waits for this allocation during startup and then performs host-local allocation of individual pod addresses from the CIDR assigned to its node.

This division of responsibility explains why C is correct. The agent consumes its assigned range and allocates endpoint addresses locally, but it does not independently choose the cluster-wide per-node PodCIDR. The Operator coordinates those ranges to prevent nodes from receiving overlapping allocations.

Options A and D describe Kubernetes host-scope IPAM rather than Cilium cluster-scope IPAM. In Kubernetes host-scope mode, the Kubernetes controller manager assigns PodCIDRs and exposes them through spec.podCIDR or spec.podCIDRs in the standard Kubernetes Node resource. Cluster-scope mode is specifically useful when Kubernetes is not configured to perform that allocation or when Cilium should control the cluster address pool.

Therefore, the managing component and resource are the Cilium Operator and CiliumNode , respectively.

Official references

Cluster-Pool IPAM ; Cluster Scope IPAM .

Study Guide topic: Installation and Configuration.

Technical explanation

Cilium tunnel mode supports VXLAN and Geneve encapsulation. In this routing model, Cilium nodes form an overlay mesh, and traffic exchanged between nodes is carried inside UDP-encapsulated packets. VXLAN is the default tunnel protocol and normally uses UDP port 8472. Geneve is the alternative and normally uses UDP port 6081. Operators select the protocol through the tunnel-protocol configuration setting, whose documented values are vxlan and geneve .

Encapsulation reduces the requirements placed on the underlying network. The underlay only needs to provide IP connectivity between the Kubernetes nodes and permit the selected UDP tunnel port. It does not need to learn or route individual PodCIDRs. Cilium also uses the tunnel metadata to carry information such as the source security identity, avoiding an additional identity lookup on the receiving node.

MPLS, OTV, STT, and EVPN are not supported values for Cilium’s tunnel-protocol setting. EVPN may be used in broader data-center network designs, and MPLS is a carrier-routing technology, but neither is a Cilium overlay encapsulation choice. Therefore, B is the only supported pair.

Official references

Cilium Routing ; System Requirements .

Study Guide topic: Architecture.

Technical explanation

In Cilium’s default policy-enforcement mode, an endpoint initially permits ingress and egress traffic. Enforcement changes independently for each direction when a policy selects that endpoint. If a selecting rule contains an ingress section, the endpoint enters default-deny mode for ingress. If a selecting rule contains an egress section, it enters default-deny mode for egress. Only traffic explicitly permitted by the applicable policy rules remains allowed in the restricted direction.

This per-direction behavior is important. An ingress-only policy does not automatically restrict egress, and an egress-only policy does not automatically restrict ingress. Options A and B reverse the relationship between the rule section and the direction being enforced. Option C incorrectly states that selection places the endpoint into default-allow mode; default allow describes the endpoint’s condition before it is selected by an enforcing policy.

Cilium also supports always and never enforcement modes. In always , enforcement applies even to endpoints not selected by policy. In never , policy enforcement is disabled. Policies can additionally use enableDefaultDeny for specialized visibility configurations, but those controls do not change the normal default behavior described in the question.

Official references

Policy Enforcement Modes .

Study Guide topic: Network Policy.

Technical explanation

The status output shows Envoy DaemonSet: disabled (using embedded mode) . In embedded mode, Cilium starts Envoy as a separate process inside the Cilium agent pod when a feature requiring Layer 7 processing—such as an L7 network policy, protocol visibility, Ingress, or Gateway API—is used. This is precisely the behavior described in option B.

The remaining choices contradict the exhibit. The Cilium DaemonSet has a desired count of three, which ordinarily indicates three schedulable nodes running Cilium, not four. Hubble Relay: OK means the aggregation component is healthy. Hubble Relay connects to the Hubble instances associated with Cilium agents and presents a multi-node API to clients such as Hubble CLI and Hubble UI. Finally, ClusterMesh: disabled rules out the claim that the installation is currently configured for Cilium Cluster Mesh service discovery and cross-cluster load balancing.

Cilium can alternatively run Envoy through the independently life-cycled cilium-envoy DaemonSet. That is not the deployment displayed here because the DaemonSet is explicitly disabled and embedded mode is reported.

Official references

Envoy deployment modes ; Hubble internals .

Study Guide topic: Architecture.

Technical explanation

Hubble provides network and security flow observability derived from Cilium’s datapath. Its flow records include source and destination workloads, namespaces, identities, ports, protocols, forwarding verdicts, and drop reasons. Hubble can therefore identify services or workloads whose connections were rejected by network policy. The Hubble CLI supports filtering by verdict, such as --verdict DROPPED , and official examples show events labeled Policy denied DROPPED . These capabilities directly answer the question posed in option D.

Option B, which is marked as correct in the supplied bank, is not supported by the stated functionality. Hubble may expose network-layer or supported application-layer metadata, including DNS and HTTP information when Layer 7 visibility is configured, but it is not a database query profiler and cannot determine which internal SQL statement ran longest merely from ordinary Hubble flows. Similarly, CPU utilization is a workload-resource metric normally obtained through Kubernetes metrics, Prometheus, or another monitoring system. Cilium BGP configuration is examined through Cilium configuration and BGP status commands rather than inferred from Hubble network flows.

The supplied answer key should therefore be corrected from B to D.

Official references

Inspecting Network Flows with the Hubble CLI ; Hubble internals .

Study Guide topic: Network Observability.

Technical explanation

For every pod managed by Cilium, the responsible node-local Cilium agent creates a CiliumEndpoint object with the same name and namespace. The object exposes endpoint state such as labels, the allocated security identity, addressing information, and effective policy. The agents also create endpoint objects for their inter-agent health endpoints. B is therefore correct.

The other responsibilities belong primarily to the Cilium Operator. The operator normally registers Cilium’s CustomResourceDefinitions, manages addresses for LoadBalancer Services when LB IPAM is active, and performs configured cluster-wide synchronization or shared-state operations. It also garbage-collects orphaned CiliumEndpoint resources when their associated pods no longer exist or have permanently completed.

The Cilium agent itself runs on each node. It responds to workload lifecycle events, creates and manages local endpoints, loads eBPF programs, applies policies, and maintains the node’s datapath. This division ensures that latency-sensitive forwarding and endpoint operations remain node-local while logically cluster-wide activities are consolidated in the operator.

Official references

CiliumEndpoint CRD , Cilium Operator Responsibilities

Study Guide topic: Cilium agent, Cilium Operator, and CiliumEndpoint lifecycle.