Hi, I’m using Cloudera CDH 5.15.1 with spark version 2.3 and Knime 4.7.2 (cdh is kerberized)
I have a workflow like this:
Connect to Hive and HDFS
Create Spark Context
Put the hive table to spark
Use Spark to Hive (using the same hive connector as previous one).
The workflow succeed. However, the new hive table is nowhere to be found.
If I use [database].[table_name], I’ll get Database not found error
Could there be anything that I should do rather than Spark to Hive to make it work?
I used to work with MapR last year with older version of Knime, All I need to do was spark to hive and it works like a charm.
So, is this the cloudera thing or the newer version of Knime is a bit different?
Or, I should do something with my clusters?
Attached is the sample workflow
testhive 12.zip (18.3 KB)