Currently, if a replica dies, it will be forever inside the cluster server. This can result in indefinite candidate states, since if enough replicas die, it will result in a candidate never achieving majority.
A "cleanup" mechanism must be found in order to address this issue.
Currently, if a replica dies, it will be forever inside the cluster server. This can result in indefinite candidate states, since if enough replicas die, it will result in a candidate never achieving majority. A "cleanup" mechanism must be found in order to address this issue.