Currently, Tajo does not provide queryable cluster information. We just can display cluster nodes through 'bin/tajo cluster' inline command.
I propose that TajoMaster should maintain a queryable virtual table where each row represents each cluster node and columns should contain hostname, resource, last ping, and so on. The virtual table should be processed by API and SQL.
Users can discovery cluster information in ad-hoc manner. Also, internal Tajo system can make good use of the virtual table for scheduling resources or displaying cluster information in WEB UI. I also expect that this feature also simplify the many parts related to cluster information.