When a cluster experiences majority failure, but there is at least one known valid copy of a member's data-dir, the cluster can be rebuilt from that copy by launching a single-node etcd cluster with the flag --force-new-cluster. This issue will track the design and development of this feature.
When a cluster experiences majority failure, but there is at least one known valid copy of a member's data-dir, the cluster can be rebuilt from that copy by launching a single-node etcd cluster with the flag
--force-new-cluster. This issue will track the design and development of this feature.