Source code

git clone https://github.com/apache/kylin.git 
# Compile 
mvn clean install -DskipTests

The environment on dev machine

Install Maven

The latest maven can be found at http://maven.apache.org/download.cgi, we create a symbol link so that mvn can be run anywhere.

cd ~
wget http://xenia.sote.hu/ftp/mirrors/www.apache.org/maven/maven-3/3.2.5/binaries/apache-maven-3.2.5-bin.tar.gz
tar -xzvf apache-maven-3.2.5-bin.tar.gz
ln -s /root/apache-maven-3.2.5/bin/mvn /usr/bin/mvn

Install Spark

Manually install the Spark binary in a local folder like /usr/local/spark. Kylin supports the community version of Spark. You can go to apache spark's official website and download spark 2.4.6.

How to Debug

There are two modes to debug source code: Debug with local metadata(recommended), or debug with Hadoop sandbox.

Configuration

Debug with local metadata 

1. Edit the properties of $KYLIN_SOURCE_DIR/examples/test_case_data/sandbox/kylin.properties

# Need to use absolute path
kylin.metadata.url=${KYLIN_SOURCE_DIR}/examples/test_case_data/sample_local
kylin.storage.url=${KYLIN_SOURCE_DIR}/examples/test_case_data/sample_local
kylin.env.zookeeper-is-local=true
kylin.env.hdfs-working-dir=file://$KYLIN_SOURCE_DIR/examples/test_case_data/sample_local
kylin.engine.spark-conf.spark.master=local
kylin.engine.spark-conf.spark.eventLog.dir=/path/to/local/dir
kylin.engine.spark-conf.spark.sql.shuffle.partitions=1
kylin.env=LOCAL

     2. Open DebugTomcat.java, start to debug. 

     3. Edit Configuration

   VM options "-Dspark.local=true"  is for query engine.

Debug with Hadoop sandbox

Local configuration must be modified to point to your Hadoop sandbox (or CLI) machine.

• In examples/test_case_data/sandbox/kylin.properties
  • Find sandbox and replace with your Hadoop hosts (if you’re using HDP sandbox, this can be skipped)
  • Find kylin.job.use-remote-cli and change it to “true” (in the code repository the default is false, which assumes running it on Hadoop CLI)
  • Find kylin.job.remote.cli.username and kylin.job.remote.cli.password, fill in the user name and password used to login Hadoop cluster for Hadoop command execution; If you’re using HDP sandbox, the default username is root and password is hadoop.
• In examples/test_case_data/sandbox
  • For each configuration XML file, find all occurrences of sandbox and sandbox.hortonworks.com, replace with your Hadoop hosts; (if you’re using HDP sandbox, this can be skipped)

An alternative to the host replacement is updating your hosts file to resolve sandbox and sandbox.hortonworks.com to the IP of your sandbox machine.

Launch Kylin Web Server

Copy server/src/main/webapp/WEB-INF to webapp/app/WEB-INF

cp -r server/src/main/webapp/WEB-INF webapp/app/WEB-INF

Download JS for Kylin web GUI. npm is part of Node.js, please search about how to install it on your OS.

cd webapp
npm install -g bower
bower --allow-root install

If you encounter a network problem when run “bower install”, you may try:

git config --global url."git://".insteadOf https://

Note, if on Windows, after installing bower, need to add the path of “bower.cmd” to system environment variable ‘PATH’, and then run:

bower.cmd --allow-root install

In IDE, launch org.apache.kylin.rest.DebugTomcat. Please set the path of the “server” module as the “Working directory”, set “kylin-server” for “Use classpath of module”, and check the “Include dependencies with ‘Provided’ scope” option in IntelliJ IDEA 2018. If you’re using IntelliJ IDEA 2017 and older, you need modify “server/kylin-server.iml” file, replace all “PROVIDED” to “COMPILE”, otherwise a “java.lang.NoClassDefFoundError: org/apache/catalina/LifecycleListener” error may be thrown.

You may also need to tune the VM options:

-Dhdp.version=2.4.0.0-169 -DSPARK_HOME=/usr/local/spark -Dkylin.hadoop.conf.dir=/workspace/kylin/examples/test_case_data/sandbox -Xms800m -Xmx800m -XX:PermSize=64M -XX:MaxNewSize=256m -XX:MaxPermSize=128m 

Also remember that if you debug with local mode, you should add a VM option for the query engine:

-Dspark.local=true

If you worked with Kerberized Hadoop Cluster, the additional VM options should be set:

-Djava.security.krb5.conf=/etc/krb5.conf -Djava.security.krb5.principal=kylin -Djava.security.krb5.keytab=/path/to/kylin/keytab

And Hadoop environment variable:

HADOOP_USER_NAME=root

By default Kylin server will listen on the 7070 port; If you want to use another port, please specify it as a parameter when run DebugTomcat.

Check Kylin Web at http://localhost:7070/kylin (user:ADMIN, password:KYLIN)

How to Package and Deploy

cd ${KYLIN_SOURCE_CODE}
# For HDP2.x
./build/script/package.sh

# For CDH5.7
./build/script/package.sh -P cdh5.7
# After finished, the package will be avaliable in the directory ${KYLIN_SOURCE_CODE}/dist/

# If running on HDP, you need to uncomment the following properties in kylin.properties
kylin.engine.spark-conf.spark.driver.extraJavaOptions=-Dhdp.version=current
kylin.engine.spark-conf.spark.yarn.am.extraJavaOptions=-Dhdp.version=current
kylin.engine.spark-conf.spark.executor.extraJavaOptions=-Dhdp.version=current