Apache Zeppelin for Denodo - User Manual

Download original document


Apache Zeppelin for Denodo is a web-based notebook. This customization of a standard distribution of Apache Zeppelin adds some new features that make it easier to use this tool with Denodo and offer a more integrated experience.

This application allows you to log in to a VDP server using the Denodo authentication and authorization infrastructure, and execute VQL sentences in your Zeppelin paragraphs.

This document contains a functional explanation of the Denodo-customized functionalities and screens of the application. For a complete reference on usage and all other features of Apache Zeppelin, please refer to the Apache Zeppelin website at https://zeppelin.apache.org/

Apache Zeppelin is developed by The Apache Software Foundation.

Installation and Configuration

Installing the notebook

This application can be downloaded from the Denodo Support Site.

Note that this application is distributed in two different versions:

      · Apache Zeppelin for Denodo - Shared Server

      · Apache Zeppelin for Denodo - Standalone

In a server that is meant to be accessed by several users in your network (concurrently or not) you should use the Shared Server version. For security reasons this version only includes the Denodo and Markdown interpreters. If you want to use additional interpreters you should install the Standalone version, meant to be used on a local machine by a single user.

The Standalone version includes the interpreters for angular, denodo, file, jdbc, md, python, sh and spark. If you need to install more interpreters you can check the Appendix III of this manual.

Once the zip file is downloaded and unzipped, you can start the application by executing bin/zeppelin.cmd on Windows, or bin/zeppelin.sh on Linux. Please note it might take up to a minute to complete the start process.

If you are using linux you need to add execution permissions executing chmod u+x bin/*.sh 

However, note that it is highly recommended to go through the rest of the “Installation and Configuration” section in this manual in order to adapt Zeppelin to your needs (e.g. connect to the right VDP server) before starting the notebook for the first time.

Configuring access to the Denodo server and database

In order to execute Denodo VQL paragraphs we have defined a customized JDBC Denodo interpreter, registered under the name denodo.

Important: please note that, when you start the Zeppelin application for the first time, all the configuration for all the interpreters is read from the interpreter/{interpretername} folder structure, and then merged together into a single conf/interpreter.json file (which contents can be afterwards modified by a Zeppelin admin user from the UI). This is standard Apache Zeppelin behaviour, and it makes it highly recommendable to configure your Denodo server access at the interpreter/denodo/interpreter-setting.json file before starting Zeppelin for the first time.

At the interpreter/denodo/interpreter-setting.json file there are two main configuration settings that you might need to change:

  • default.url containing the JDBC URI to be used to connect to the desired VDP server. Default value is jdbc:vdb://localhost:9999/?noAuth=true

  • zeppelin.jdbc.denodo.database containing the name of the default database to be used for paragraph execution (note that paragraphs can later specify a different database to be executed on). Please note that the VDP users used for accessing Zeppelin need to have connection privileges on the database configured here. Default value is admin.

After the first execution, the admin user will be responsible for making the configuration changes in the interpreter section of the application.

Configuring authentication

The authentication mechanisms of the Denodo-customized Zeppelin application are integrated with the VDP server being accessed.

Authentication is always performed on a VDP server, and the application can use Denodo SSO, Kerberos-based or basic (user+password) authentication. By default only basic authentication is enabled. You can modify this behavior to configure the allowed authentication methods by changing the zeppelin.denodo.login.types property at the  conf/zeppelin-site.xml file. Possible values for this property ​​are BASIC, KERBEROS,DENODOSSO and ALL. If the property is not defined, all types of login are enabled. DENODOSSO authentication is only available on Denodo 8 and later versions.

When all authentication methods are enabled, Denodo SSO will have the priority followed by KERBEROS and BASIC. If the Denodo SSO configuration is valid zeppelin will authenticate through Denodo SSO and the configured identity provider. If Denodo SSO is not configured or the configuration is not valid, the next authentication method is Kerberos. So if your browser is configured so that it is able to obtain a valid Kerberos token from the Operating System infrastructure, the application will try to log you in using Kerberos. This will happen even if your Denodo installation is not configured for using Kerberos, so this might lead to inability to log in if you have KERBEROS or ALL as configured login type, your browser is integrated with your OSs Kerberos infrastructure, and your VDP is not.

If Zeppelin is configured to use ALL login types but your DenodoSSO application is not working the zeppelin application tries to use Kerberos and if your browser is not integrated with a Kerberos infrastructure, the application will default to basic authentication using a username and password. 

To configure DenodoSSO in zeppelin you need to check the zeppelin.denodo.login.ssourl property at the  conf/zeppelin-site.xml file. This is the URL of the Denodo SSO application, by default the value is http://localhost:9090/sso but maybe it is necessary to change the host and/or the port.

For the configuration of DenodoSSO it is important to configure the Denodo Security Token Server. This configuration is more detailed in the DenodoSSO documentation. The configuration files you need to check are the SSOConfiguration.properties, SSOTokenConfiguration.properties and the tokenJavaKeyStore.jks. It is also important to correctly configure the redirection URL’s in  the identity provider, make sure that the host of the sso URL configured in zeppelin is the same as in the redirection URL’s configured in the identity provider. You must configure in vdp the privileges of the role of the Denodo Security Token Server. This role needs to have access to the databases that need to be accessed from Zeppelin.

Also note that the implementation of SPNEGO in Microsoft Web browsers negotiates between Kerberos and NTLM mechanisms. This means that if both browser and server support Kerberos, this will be used, but if for some reason Kerberos is not possible the browser will try NTLM. You therefore might be getting NTLM tokens sent to the Zeppelin server because the browser is unable to authenticate using Kerberos. This behaviour occurs with Chrome, Edge or Internet Explorer because it uses OS settings. In the case of Firefox, the browser has its own configuration and doesn’t use NTLM when Kerberos is not configured. If the browser uses the NTLM token we will get an error from the zeppelin authentication mechanism. This error happens with Chrome or Internet Explorer and the solution is to change the zeppelin.denodo.login.types property to BASIC or KERBEROS or use Firefox.

Another thing one has to bear in mind is the logout. In some browsers like firefox the basic authentication doesn't expire and if you try to log out the browser sends the authentication header again and logs in again automatically.

In the scope of this application, Zeppelin users can adopt two different roles: “admin user” and “data scientist user”. The “admin user” can modify specific configurations like create and modify interpreters. The “data scientist user” can execute notes and jobs. The admin user can also execute paragraphs and jobs. For a user to be admin, they must have the assignprivileges role in Denodo. This role will be assigned in the role management section of the Denodo administration tool.

Notebook storage

Apache Zeppelin has a pluggable notebook storage mechanism controlled by zeppelin.notebook.storage configuration option with multiple implementations.

By default, Zeppelin stores the information of the notebooks and paragraphs in json files inside the folder /notebook. We have decided that this information is significantly less exposed if it is stored in a derby database. For this we have added a new NotebookRepo org.apache.zeppelin.notebook.repo.DerbyDBNotebookRepo that will be configured in the file /conf/zeppelin-site.xml using the parameter zeppelin.notebook.storage.




  <description>DerbyDB notebook persistence layer implementation</description>


All other notebook storage implementations can be used at the same time as the new DerbyDBNotebookRepo implementation. You only need to add them in the configuration parameter, separated by commas.

It is important to keep in mind that if other implementations are activated, the notebooks will be stored using both implementations. And therefore, the json files will continue to be exposed on the system where they are stored.

Accessing the notebook from a browser

Once started, the notebook can be accessed at http://localhost:8080 (if needed, replace localhost with the name of the machine it has been installed on).

Modify the configuration in conf/zeppelin-site.xml if you need to change the port Zeppelin starts at, or want to configure Zeppelin to be accessed via HTTPS.

Paragraph execution

We have made some changes in the paragraph execution for the Denodo Interpreter. In this interpreter we create a connection with Denodo in each paragraph and we allow to specify the database to connect. By default we will use the database defined in the zeppelin.jdbc.denodo.database property of the interpreter configuration but if we want to connect with a different database we can write the database name concatenated with the interpreter name. The format is %denodo%DATABASE_NAME.

Appendix I: Configuring browsers for Kerberos


By default Firefox does not enable SPNEGO authentication, and consequently Kerberos. It has to be enabled manually therefore.

Enable Kerberos Authentication
  1. Go to about:config in the address bar in Firefox
  2. Click “I'll be careful, I promise” when warned about changing advanced settings
  3. Enter negotiate in the Search box
  4. Set value of the network.negotiate-auth.trusted-uris to your domain name
Enable Kerberos Delegation
  1. Go to about:config in the address bar in Firefox
  2. Click “I'll be careful, I promise” when warned about changing advanced settings
  3. Enter negotiate in the Search box
  4. Set value of the network.negotiate-auth.delegation-uris to your domain name

Restart browser and check that everything works.

Internet Explorer

Enable Kerberos Authentication
  1. Click Tools -> Internet Options
  2. Advanced tab
  3. Enable checkbox for Enable Integrated Windows Authentication

Enable Kerberos Delegation
  1. Click Tools -> Internet Options
  2. Security tab -> Local intranet -> Sites

  1. Add the site in question.

Restart browser and see if everything works.


Chrome in Windows will use the Internet Explorer settings, so configure them within Internet Explorer's Tools -> Internet Options dialog as explained in the previous section.

Appendix II: Updating the Denodo JDBC Driver

The Denodo-customized Zeppelin distribution is already provided with the Denodo VDP JDBC Driver (denodo-vdp-jdbcdriver.jar) included in the interpreter/denodo folder. You may need to update this driver to match any update you have installed on the Denodo VDP server

To update de Denodo VDP JDBC Driver you must take the driver included in the lib/extensions/jdbc-drivers/vdp-VERSION directory of your Denodo platform and copy into the Zeppelin interpreter/denodo folder.

Appendix III: Installing additional interpreters

The Apache Zeppelin for Denodo Standalone distribution is provided with some pre installed interpreters (angular, denodo, file, jdbc, md, python, sh, spark). The original distribution provides a Linux script to install interpreters and we add a new script to install interpreters in Windows.

The new script is install-interpreter.cmd and the execution is the same as Linux.

Install all community managed interpreters

.\bin\install-interpreter.cmd --all

Install specific interpreters

.\bin\install-interpreter.cmd --name md,shell,jdbc,python

Get list of community managed interpreters

.\bin\install-interpreter.cmd --list

Appendix IV: Using the Python interpreter in Windows

The Python interpreter is already included in the Apache Zeppelin for Denodo - Standalone distribution. This interpreter is compatible up to Python version 3.7.

To use it in Windows, it will be necessary to have the directory "C:\tmp" and that the user who is running zeppelin has permissions on this directory.