Using Hive on Tez with HDP2.5 sandbox. I submit a query (select * from ...) from beeline and kill it immediately (after ~1s). The application is killed, and the following log line appears: Could not connect to AM, killing session via YARN, sessionName=HIVE-xxx, applicationId=application_xxx My problem is, that another container is started immediately with the same sessionName which begins to execute my previously cancelled query. This does not happen is I kill the application after 15s, in that case the application's final status is SUCCEDED (killedDags=1). ... View more

I'd like to bulk insert records from my client into a Hive table via JDBC (to be clear: the data is not on the Hadoop cluster, but on the local file system of the client. I can't upload it to the cluster or to HDFS.). The best I could achieve so far is calling insert into...values using hive on tez, which gives ~10000 records/sec throughput, but this is still very poor. I've also checked Hive Streaming Data Ingest, but for that I need to connect to the Hive Metastore, which is not an option in my case. Unfortunately I can connect only to the hiveserver2 port (10000). Any other ideas? ... View more

I'd like to create a custom Ambari service that modifies the HIVE_CLASSPATH env variable and the mapreduce.application.classpath Hadoop property (i. e. I want that Hive and Hadoop use these extended classpaths). Is it possible to do that in Ambari? Where can I find some examples about that? Thanks ... View more