preconfiguration:
replic: 1:1
shard: 3
stored 30 indexs, each index include about 4GB data. total data is about 120GB
operation:
1. create 2 nodes in one cluster
2. do search with kibana
observed result:
1. the search task is not Concurrent. it first do searching from one node, then go to another node after several minutes
2. the searching ability of 2 nodes is worse than 1 nodes
3. the issue happens in ES5.0.1 and ES5.1.2, not in ES2.3.2