In a managed public cloud Kubernetes service, which component is most closely associated with issues like a LoadBalancer being stuck in a pending state?

• A. kube-scheduler
• B. etcd
• C. Cloud Controller Manager
• D. kube-proxy

Answer: (C)

The key concept is the cloud provider integration that provisions external cloud resources for Kubernetes. When you create a service of type LoadBalancer in a managed public cloud, the component that talks to the cloud provider to actually create and configure the external load balancer is the Cloud Controller Manager. It runs cloud-specific controllers that manage resources like external load balancers, routes, and instances. If the LoadBalancer stays in a pending state, it often points to an issue with this cloud integration—CCM not communicating with the cloud API, credentials problems, or the CCM not running or syncing correctly.

Other components have different duties. The scheduler decides which node runs the service pods, not anything to do with provisioning cloud resources. The etcd datastore holds cluster state, but not cloud resources. The kube-proxy handles internal service traffic routing within the cluster, using iptables or IPVS, but it does not create or manage external load balancers.