Improvements to a Roll-Back Mechanism for Asynchronous Checkpointing and Recovery
Monika Kapus-Kolar · 2009
We indicate the existence of a logical flaw in a recently published recovery algorithm for distributed systems and suggest a correction. We also improve the associated communication protocol, the prevention of multiple concurrent instantiations of the algorithm, the handling of obsolete messages and the organization of the stable storing of the relevant data. Povzetek: Opozarjamo na logično napako v pred kratkim objavljenem algoritmu za okrevanje v porazdeljenih sistemih in predlagamo popravek.