Abstract: Some embodiments provide a system and method associated with improving resilience for a non-high availability service provided by a multi-node cluster (e.g., a bosh-based database service). A processor of a node in the cluster may determine that a lifecycle event has occurred (e.g., an initial deployment, failover, or deployment update). Responsive to the lifecycle event, the processor may store information about a role of a master node and timestamp information (e.g., on persistent storage). The processor may subsequently determine that a cluster deployment crash has occurred for the service. Timestamp information associated with multiple nodes may then be compared to establish one of the nodes as the master node. For example, the most recent timestamp may be used to establish the master node.

Abstract: Some embodiments provide a system and method associated with improving resilience for a non-high availability service provided by a multi-node cluster (e.g., a bosh-based database service). A processor of a node in the cluster may determine that a lifecycle event has occurred (e.g., an initial deployment, failover, or deployment update). Responsive to the lifecycle event, the processor may store information about a role of a master node and timestamp information (e.g., on persistent storage). The processor may subsequently determine that a cluster deployment crash has occurred for the service. Timestamp information associated with multiple nodes may then be compared to establish one of the nodes as the master node. For example, the most recent timestamp may be used to establish the master node.