Introduction

In Hue-2745 v3.10, add JDBC support like Phoenix, Kylin, Redshift, Solr Parallel SQL, …

However, there isn’t any manual to use with Kylin.

Pre-requisites

Build a cube sample of Kylin with: Quick Start with Sample Cube, will be enough.

You can check:

Used Software:

Install Hue

If you have Hue installed, you can skip this step.

To install Hue on Ubuntu 16.04 LTS. The official Instructions didn’t work but this works fine:

There isn’t any binary package thus pre-requisites must be installed and compile with the command make

sudo apt-get install --fix-missing -q -y \
    git \
    ant \
    gcc \
    g++ \
    libkrb5-dev \
    libmysqlclient-dev \
    libssl-dev \
    libsasl2-dev \
    libsasl2-modules-gssapi-mit \
    libsqlite3-dev \
    libtidy-0.99-0 \
    libxml2-dev \
    libxslt-dev \
    libffi-dev \
    make \
    maven \
    libldap2-dev \
    python-dev \
    python-setuptools \
    libgmp3-dev \
    libz-dev

Download and Compile:

git clone https://github.com/cloudera/hue.git
    cd hue
    make apps

Start and connect to Hue:

build/env/bin/hue runserver_plus localhost:8888
  • runserver_plus: is like runserver with debugger
  • localIP: Port, usually Hue uses 8888

The output must be similar to:

Connect using your browser: http://localhost:8888

Important: The first time that you connect to hue, you set Login / Pass for admin

We will use Hue / Hue as login / pass

Issue 1: Could not create home directory

It is a permission problem of your current user, you can use: sudo to start Hue

Issue 2: Could not connect to …

If Hue’s code had been downloaded from Git, Hive connection is active but not configured → skip this message

Issue 3: Address already in use

The port is in use or you have a Hive process running already

You can use *ps -ef grep hue*, to find the PID and kill

Configure Hue for Apache Kylin

The purpose is to add a snipped in a notebook with Kylin queries

References:
* Custom SQL Databases
* Manual: Kylin JDBC Driver
* GitHub: Kylin JDBC Driver

Register JDBC Driver

  1. To find the JAR Class for the JDBC Connector

From Kylin Download
Choose Binary and the correct version of Kylin and HBase

Download & Unpack: in ./lib:

  1. Place this JAR in Java ClassPATH using .bashrc

check the actual value: alt text

check the permission for this file (must be accessible to you):

  1. Add this new interface to Hue.ini

Where is the hue.ini ?

  • If the code is downloaded from Git: UnzipPath/desktop/conf/pseudo-distributed.ini

    (I shared my INI file in GitHub).

  • If you are using Cloudera: you must use Advanced Configuration Snippet

  • Other: find your actual hue.ini

Add these lines in [[interpreters]]

[[[kylin]]]
    name=kylin JDBC
    interface=jdbc
    options='{"url": "jdbc:kylin://172.17.0.2:7070/learn_kylin","driver": "org.apache.kylin.jdbc.Driver", "user": "ADMIN", "password": "KYLIN"}'
  1. Try to Start Hue and connect just like in ‘Start and connect’

TIP: One JDBC Source for each project is need

Register without a password, it can do use this other format:

options='{"url": "jdbc:kylin://172.17.0.2:7070/learn_kylin","driver": "org.apache.kylin.jdbc.Driver"}'

And when you open the Notebook, Hue prompts this:

Issue 1: Hue can’t Start

If you see this when you connect to Hue ( http://localhost:8888 ):

Go to the last line alt text

And launch Python Interpreter (see console icon on the right):

In this case: I’ve forgotten to close “ after learn_kylin

Issue 2: Password Prompting

In Hue 3.11 there is a bug Hue 4716

In Hue 3.10 with Kylin, I don’t have any problem :)

Test query example

Add Kylin JDBC as source in the Kylin’s notebook:

alt text > alt text > alt text > alt text

Write a query, like this:

select part_dt, sum(price) as total_selled, count(distinct seller_id) as sellers from kylin_sales group by part_dt order by part_dt

And Execute with: alt text

Congratulations !!! you are connected to Hue with Kylin

Issue 1: No suitable driver found for jdbc:kylin

There is a bug, not solved since 27 Aug 2016, nor in 3.10 and 3.11, but the solution is very easy:

Link:
You only need to change 3 lines in */desktop/libs/librdbms/src/librdbms/jdbc.py*

Limits

In Hue 3.10 and 3.11
* Auto-complete doesn’t work on JDBC interfaces
* Max 1000 records. There is a limitation on JDBC interfaces, because Hue does not support result pagination Hue 3419.

Future Work

Dashboards
There is an amazing feature of Hue: Search Dasboards / Dynamic Dashboards. You can ‘play’ with this Demo On-line. But this only works with SolR.

There is a JIRA to solve this: Hue 3228, is in roadmap for 4.1. Check Hue MailListMailList and add Dashboards to JDBC connections.

Chart & Dynamic Filter
Nowadays, it isn’t compatible, you only can work with Grid.

DB Query
DB Query does not yet support JDBC.