[TASK][MEDIUM] Allow returning custom response for GetInfo request

[pan3793]

Code of Conduct

• I agree to follow this project's Code of Conduct

Search before creating

• I have searched in the task list and found no similar tasks.

Mentor

• I have sufficient knowledge and experience of this task, and I volunteer to be the mentor of this task to guide contributors to complete the task.

Skill requirements

• Basic knowledge on Scala Programming Language
• Basic knowledge on Network programming and Apache Thrift RPC framework
• Basic knowledge on ODBC and Power BI / Tableau

Background and Goals

Kyuubi implements the Hive-compatible Thrift-based API, just as Impala and Spark Thrift Server do, so technically, the Hive JDBC/ODBC clients based on Thrift-API should work smoothly with Apache Kyuubi.

Unfortunately, we found some clients verify the GetInfo results and may reject the connection from an unrecognized Server. See details at #3032

After #3122, it allows a return of either SERVER’s information or ENGINE’s information in the GetInfo response, but it is not always sufficient, we may want to make it configurable so that the user can configure Kyuubi to return any information they want, to make it’s easy to impersonate any kind of Server to allow ODBC clients like PowerBI and Tableau to connect.

• Make ODBC-based clients support Apache Kyuubi smoothly
• Supply docs for how to use Power BI and Tableau connect to Apache Kyuubi

Implementation steps

Currently, the value candidates of kyuubi.server.info.provider are SERVER and ENGINE, we can introduce a new option CUSTOM with additional configuration to allow users to configure each property of GetInfo response.

• Make Kyuubi Server support return a custom GetInfo response
• Supply docs for how to use Power BI connect to Apache Kyuubi
• Supply docs for how to use Tableau connect to Apache Kyuubi

Additional context

Introduction of #6232

[BruceWong96]

Hi @pan3793, I am very interested in this task, can you assign it to me ? Thanks.

[pan3793]

@BruceWong96 thanks, do you have an estimated deadline for this task?

[BruceWong96]

@BruceWong96 thanks, do you have an estimated deadline for this task?

If everything goes well, I will do my best to finish by November 15th.

[BruceWong96]

Hello, @pan3793
Sorry for missing the time, but I realized when I was writing the code that I must have misunderstood the work plan
How can the getInfo interface customize what it returns?
In what form do we customize key and value?
Do we provide an abstract interface for the user to implement?
Because I'm not sure what kind of information the user wants to request.
I need some more detailed guidance.

Thank you.

[pan3793]

do we customize key and value?

I think so. define a configuration prefix, and the user could supply a serial of key-values then.

Do we provide an abstract interface for the user to implement?

Not yet. At least it is not covered by this task's scope. But I'm open to discuss the details if you think it's necessary or it's useful in some cases

I'm not sure what kind of information the user wants to request.

There were several close-source ODBC driver implementations, which is compatible with the Hive thrift protocol, but may not be used to connect to Kyuubi, because they can not recognize the information returned from GetInfo

[BruceWong96]

There were several close-source ODBC driver implementations, which is compatible with the Hive thrift protocol, but may not be used to connect to Kyuubi, because they can not recognize the information returned from GetInfo

Function getInfo

About some details, as above code.
The closed-source ODBC driver carries the request code when it sends the request getInfo. The parameter type of getInfo is TGetInfoType. Should the client custom key also be TGetInfoType?

Do you have any more ideas about the client's request parameters?

Looking forward to your reply.

[pan3793]

Should the client custom key also be TGetInfoType?

Yea, the client MUST respect the Hive-defined thrift protocol.