This work discusses the problem of correctly instantiating and deploying monitors that can accurately detect deadlocks in a distributed message-passing system.

We define formal notions of monitoring correctness, and we discuss the nuances of the problem — which depend on how the components of a distributed system may behave and deadlock, and how the monitors can observe the communications between such components.

We also outline our ongoing research on the topic, and our current results, which cover a relevant class of RPC-based systems.