A method for detecting distributed concurrency errors in a distributed cloud computing system includes tracing operations that access objects in functions involving inter-process messaging, applying a set of happens-before rules to the traced operations. Analyzing the traced operations to identify concurrent operations that access a common object to generate a list of potential distributed concurrency errors (DCbugs). Pruning the list of DCbugs to remove DCbugs having only local effect and that do not generate run-time errors.