Hi, I am trying to run a hive process using oozie and it is taking too long to process. The normal hive query takes about 35 min and when scheduled with oozie it takes 2 hours 45 minutes. I checked the logs and found out that there are 75000 splits which the hive query is trying to process. When run from hive prompt or using a shell script it groups the 75000 splits and makes it in to 450 splits. This grouping doesnt happen when I use hive with oozie. I set tez.job.queuename property to prod but that doesnt resolve the problem. Can some one help me to group these splits.
I digged in a little more and found that the execution engine when I run through CLI is tez and when I run through oozie is hive. So I added a property hive.execution.engine as shown below. I also added one more property oozie.hive.details which points to hdfs location as shown below - hive-config.xml. I ran with these changes but still I am unable to set the default engine as "tez"