Migrate session_length/daily from Oozie to Airflow
3 changes compared to the original oozie job:
- host_properties UDF is not needed anymore, the data is precomputed
-
Order by 1
in the window fct has been added because the job is now run by Spark in place of Hive (See hql file. Link below) - Use snappy compression
Linked with:
Edited by Aqu