Hadoop/Impala with Windows Authentication

database
bigdata
genericknimenodes
#1

Hi there!

I need to connect to an Hadoop table throught Impala Connector.
I know that the server is protected with Kerberos, but I want to bypass it.

I also use Microsoft Power BI and I have reached the goal using the Windows Authentication.

Can someone help me to “translate” the following info to configure the Impala Connector in Knime? the following configuration works perfectly in Microsoft Power BI only using the host name and setting the Windows current credentials.

In Knime I have only added the Cloudera ODBC driver:

I don’t know if I have to (and how to) configure this page:

  1. Is it correct the ODBC driver? Have I to install the JDBC one?
  2. How to configure the Impala Connector node to use the Windows credential Authentication?

Thanks, Andrea

0 Likes

#2

Hi @aesposito

currently there is no good way in KNIME to use the Kerberos ticket that your machine obtained during Windows login. This is an unfortunate limitation of the Java platform that KNIME is built upon.

Have you considered our blog post on this topic?
https://www.knime.com/blog/speaking-kerberos-with-knime-big-data-extensions
It walks you through the necessary steps of getting a Kerberos ticket that KNIME can read and then setting up the connector nodes.

To answer your more specific questions:

  1. Is it correct the ODBC driver? Have I to install the JDBC one?

The Impala Connector node does not support ODBC drivers. It supports two different JDBC drivers:

  1. The built-in JBDC driver for Hive/Impala (here you don’t have to install any drivers yourself. The driver comes with the node). This is recommended.
  2. Simba-based JDBC drivers for Impala, such as the one provided by Cloudera:
    https://www.cloudera.com/downloads/connectors/impala/jdbc/2-6-9.html
  1. How to configure the Impala Connector node to use the Windows credential Authentication?

As I wrote above, currently there is no good way in KNIME to use the Kerberos ticket that your machine obtained during Windows login. You have to follow the steps from the blog post to get a new ticket that is stored in a file cache.

Best,
Björn

0 Likes

#3

Hi Bjorn,
Thanks for the answer!

I wrote the topic only because I wanted to investigate about different ways in approaching to the problem!

By your reply I have understood that:

  • It is recommended to use the built-in JDBC driver (The procedure File > Preferences > KNIME > Databases > Add file/directory it is not necessary, right? Maybe do I have to add file/directory from the KNIME path?);
  • I cannot (for the moment) bypass Kerberos

Thanks a lot,
Andrea

0 Likes

#4

Hi @aesposito

It is recommended to use the built-in JDBC driver (The procedure File > Preferences > KNIME > Databases > Add file/directory it is not necessary, right? Maybe do I have to add file/directory from the KNIME path?);

Yes, it is recommended to use the builtin driver. Going through File > Preferences > KNIME > Databases > Add file/directory it is not necessary in this case.

I cannot (for the moment) bypass Kerberos

No, this is not possible.

0 Likes

#5

Thanks Bjorn.

Have a nice day,
Andrea

0 Likes

closed #6

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.

0 Likes