We have started using HDP 3.0 and we are connecting to hive through beeline but we are encountering concurrency issues. All sessions are using the hive user to connect and we are finding that when a second session tries to execute a query it remains in a pending status until the second session is closed. Any ideas where to start looking to determine why two or more sessions using the same user cannot execute queries simultaneously?
Here is the configuration details of our HDP 3.0 cluster:
1. What is concurrency set for LLAP, how may TEZ AM can you see in the resource Manager UI ?
2. Can you kindly look at the TEZ UI and graphana LLAP to find if the first query is occupying all the LLAP executors.
As your LLAP size is quite small I believe even premepetion is not helping to launch two jobs simulataneously.