This article mainly introduces how to realize the observability of DNS faults and the diagnosis of difficult problems in Kubernetes clusters.