SPLIT BRAIN RESISTANT FAILOVER IN HIGH AVAILABILITY CLUSTERS

摘要

The present invention relates to providing failover between cluster machines of a master candidate group while avoiding split brain phenomenon between cluster machines. In accordance with the present invention, a method is provided for synchronously replicating states to provide failover between nodes in a high availability cluster. Nodes in a cluster have machines, such as computers, servers, and other devices that serve network clients. According to this method, the number m of master candidates and the number n of quorum machines are identified among the machines belonging to the cluster. To provide failover, m≥2. For quorum to be meaningful, it must be n≥3 and odd.