diff --git a/examples/sqoop-parquet-full-load/tables.yml b/examples/sqoop-parquet-full-load/tables.yml index efbea19..77944a0 100644 --- a/examples/sqoop-parquet-full-load/tables.yml +++ b/examples/sqoop-parquet-full-load/tables.yml @@ -15,10 +15,10 @@ source_database: cmd: "mysql -P 3306 -uroot -pcloudera -h mysql <" staging_database: name: "default" # Staging database name. - path: "{{ hdfs_basedir }}/staging" # Staging database HDFS path + path: "{{ hdfs_basedir }}/clean" # Staging database HDFS path raw_database: name: "default" # Result database - path: "{{ hdfs_basedir }}/result" # Result database HDFS path + path: "{{ hdfs_basedir }}/raw" # Result database HDFS path final_database: name: "default" # Result database tables: @@ -31,7 +31,7 @@ tables: source: name: "employees_kudu" # Source table name destination: - name: "employees/kudu.test" # Destination (Impala) table name + name: "employees_kudu_test" # Destination (Impala) table name. Typically this is the same as the source table split_by_column: "emp_no" # Sqoop split by column (--split-by) primary_keys: # List of primary keys - emp_no