A case study on parametric verification of failure detectors

Partial synchrony is a model of computation in many distributed algorithms and modern blockchains. These are typically parameterized the number participants, their correctness requires existence bounds on message delays relative speed processes after reaching Global Stabilization Time. characteristics make partially synchronous processes, parametric time bounds, which render automated verification challenging. In this paper, we present case study formal both safety liveness Chandra Toueg failure detector that based partial synchrony. To end, first introduce formalize class symmetric point-to-point contains detector. Second, show these have cutoff, cutoff results hold three models computation: synchrony, asynchrony, As result, one can verify them by checking small instances, but problem stays time. Next, specify assumptions frameworks: TLA+, IVy, counter automata. Importantly, tune our modeling to use strength each method: (1) We using counters encode buffers with automata, (2) first-order relations (3) approaches TLA+. By running tools for TLA+ demonstrate fixed bounds. prove arbitrary Moreover, how reducing verification. Thus, properties verified developing inductive invariants IVy.