Mutation Testing for Container Orchestration Declarative Configurations: An Empirical Evaluation on Kubernetes

As cloud-native software architectures increasingly rely on declarative Infrastructure as Code (IaC) paradigms, ensuring the correctness and resilience of container orchestration configurations has become critical. Misconfigurations in Kubernetes manifests frequently lead to severe runtime outages, security vulnerabilities, and deployment failures. While traditional software engineering employs mutation testing to evaluate the effectiveness of test suites, its application to declarative container orchestration remains largely unexplored. In this paper, we propose a domain-specific mutation testing framework tailored for Kubernetes declarative configurations. We define a comprehensive taxonomy of mutation operators targeting pod resource specifications, health probes, networking definitions, deployment strategies, and security contexts. We conduct an empirical evaluation across 25 production-grade open-source cloud repositories, generating and executing over 4,500 mutants against existing static linters and dynamic validation suites. Our findings reveal that conventional static analysis tools achieve an average mutation score of only 38.4%, whereas dynamic cluster integration testing reaches 64.2%. Combining static analysis with automated dynamic assertion suites yields an overall mutation score of 88.7%. These results highlight significant diagnostic blind spots in modern DevOps pipelines and demonstrate the viability of mutation testing for elevating the reliability of cloud deployment manifests.