Hi,

I'm trying to use JdbcHook and noticed it requires to specify the
whole JDBC URL in the "host" column. An example that works is as
follows:

mysql> SELECT * FROM connection WHERE conn_id='jdbc_default'\G
*************************** 1. row ***************************
                id: 29
           conn_id: jdbc_default
         conn_type: jdbc
              host: jdbc:hive2://172.17.0.2:10000/default
            schema: NULL
             login: hive
          password: NULL
              port: NULL
             extra:
{"extra__jdbc__drv_path":"/path/to/jar1,/path/to/jar2",
"extra__jdbc__drv_clsname":"org.apache.hive.jdbc.HiveDriver"}
      is_encrypted: 0
is_extra_encrypted: 0

But it's not intuitive and not consistent with other hooks. I think
users will expect the following manner:

           conn_id: jdbc_default
         conn_type: jdbc:hive2
              host: 172.17.0.2
            schema: default
             login: hive
          password: NULL
              port: 10000

So I'd like to fix JdbcHook as the latter, but I understand it brakes
backward compatibility.
Is this change acceptable? And if it's OK, what Airflow version should
I target (maybe 2.0)?

Kengo Seki <sek...@apache.org>

Reply via email to