I'm trying to evaluate Apache Kylin on HDP 2.6 for a PoC at my company. In the sample tutorial, after lading the tables from Hive , a MR job should calculates the cardinality of the the table. But nothing is written and I can't find any MR jobs in my cluster doing this task.
When I launch the build actionfor the cube I get stuck at the first phase "Create Intermediate Flat Hive Table".
Anyone can help me on those problems ?
Thank you in advance.
Hello, thx @Dave Russell. Yes, it was not very clear. I'm following the tutorial on the Kylin doc http://kylin.apache.org/docs20/tutorial/create_cube.html. Precisely point II.6, where there is the cardinality calculation.
For my second problem, after some investigation, Kylin is not stuck as I was saying but it takes more than 45 minutes to insert the data in the HIVE table. And if I do it manually directly in hive it takes 5 minutes top
Thank you all.
The cardinality job will be started in the backend, there is no GUI to track it now. As I know if the source table is a Hive view, the job will fail because it uses Hive HCatalog API to read the table, while view is not supported.
If it got stuck in the first step "create hive intermediate table", sometimes it indicates no enough resource on YARN. You can copy the same HQL into hive command line, it should have the same behavior because Kylin is running hive command in shell.
Hope this can help.