Eventual Leader Election in the Crash-Recovery Failure Model

Abstract

Unreliable failure detectors provide information about process failures. A particular failure detector called Omega has been shown to be the weakest for solving consensus with a majority of correct processes. This work addresses the implementation of Omega in the crash-recovery failure model. Firstly, the definition of Omega is adapted to that model, assuming that processes do not use stable storage. After that, an algorithm implementing Omega under some weak assumptions on communication reliability and synchrony is proposed.