A Semi-Automatic Correctness Proof Procedure applied to Stoller's Leader Election Algorithm
Rapport, 2008

In 1997, Stoller presented a leader election algorithm for a synchronous system with crash failures. The algorithm is an adaptation of Garcia-Molina's Bully Algorithm that uses failure detectors instead of explicit timeouts. Since the characteristics of the algorithm closely resemble the Bully Algorithm Stoller does not give a formal correctness proof. However, although the algorithms appear similar, there are non-trivial differences. The differences make it unclear if the original proof, by Garcia-Molina, actually carries over as indicated by Stoller. In this document we formalize the leader election algorithm using first-order logic, and prove its correctness with respect to the obvious safety property; it should not be possible to elect two different leaders at the same time.

Författare

Hans Svensson

Chalmers, Data- och informationsteknik, Datavetenskap

Ämneskategorier

Programvaruteknik

Datavetenskap (datalogi)

Technical report - Department of Computer Science and Engineering, Chalmers University of Technology and Göteborg University: 2008:7