This paper focuses on runtime verification of distributed systems in the partial synchronous model, where a clock synchronization algorithm ensures a bound on maximum clock skew among all processes. We introduce two centralized monitoring technique where the specification in the linear temporal logic (LTL) is either represented by a deterministic finite automaton, or, we use a progression-based formula rewitting technique to reduce the distributed runtime verification problem to an SMT problem. We report on rigorous synthetic, as well as real-world case studies involving Cassandra (a non-SQL database management system) and data available from RACE (Runtime for Airspace Concept Evaluation) by NASA. We show that both our automata-based as well as our progression-based approached are effective in evaluating all the possible verdicts given a distributed computation with the progression-based approach having less overhead in general.
