I am planning to use Elasticsearch within a single-node cluster. The other servers on our network are tasked for different resource-intensive tasks so adding additional nodes may not be feasible. If I configure path.data to multiple paths spread out over different physical disks (say 2 for example), then configure a new index with number_of_shards : 1 and number_of_replicas : 1, will elasticsearch automatically separate the primary shard and replica shard out over different disks on the same node, or must this allocation be configured? I would like to do this to at least provide some redundancy against hard drive failure (i.e. be RAID-like) if we cannot run with more than a single-node cluster.
Separately, even if the primary and replica shards are spread between different disks, can the overall cluster status ever be green?