Bugs in popular distributed protocol implementations have been the source of many downtimes in popular internet services. We describe a randomized testing approach for distributed protocol implementations based on reinforcement learning. Since the natural reward structure is very sparse, the key to successful exploration in reinforcement learning is reward augmentation. We show two different techniques that build on one another. First, we provide a decaying exploration bonus based on the discovery of new states—the reward decays as the same state is visited multiple times. The exploration bonus captures the intuition from coverage-guided fuzzing of prioritizing new coverage points; in contrast to other schemes, we show that taking the maximum of the bonus and the Q-value leads to more effective exploration. Second, we provide waypoints to the algorithm as a sequence of predicates that capture interesting semantic scenarios. Waypoints exploit designer insight about the protocol and guide the exploration to “interesting” parts of the state space. Our reward structure ensures that new episodes can reliably get to deep interesting states even without execution caching. We have implemented our algorithm in Go. Our evaluation on three large benchmarks (RedisRaft, Etcd, and RSL) shows that our algorithm can significantly outperform baseline approaches in terms of coverage and bug finding.

Cloud computing has radically changed the way organizations operate their Software by allowing them to achieve high availability of services at affordable cost. Containerized microservices is an enabling technology for this change, and advanced container orchestration platforms such as Kubernetes are used for service management. Despite the flourishing ecosystem of monitoring tools for such orchestration platforms, service management is still mainly a manual effort. The modeling of cloud computing systems is an essential step towards automatic management, but the modeling of cloud systems of such complexity remains challenging and, as yet, unaddressed. In fact modeling resource consumption will be a key to comparing the outcome of possible deployment scenarios. This paper considers how to derive resource models for cloud systems empirically. We do so based on models of deployed services in a formal modeling language with explicit CPU and memory resources; once the adherence to the real system is good enough, formal properties can be verified in the model. Targeting a likely microservices application, we present a model of Kubernetes developed in Real-Time ABS. We report on leveraging data collected empirically from small deployments to simulate the execution of higher intensity scenarios on larger deployments. We discuss the challenges and limitations that arise from this approach, and identify constraints under which we obtain satisfactory accuracy.