The vulnerability stems from improper validation of cluster size input during service discovery. The core issue occurs when a negative cluster size is used in array indexing operations. The etcdserver component handles cluster initialization and member discovery, making Server.startCluster a prime candidate as it would be responsible for processing cluster configuration parameters. The panic occurs because negative values would cause invalid index calculations when initializing cluster member lists or performing service discovery lookups. While exact line numbers aren't available, the described behavior and component structure strongly implicate the cluster initialization path in etcdserver.