Apache Kylin : Analytical Data Warehouse for Big Data
Welcome to Kylin Wiki.
|email@example.com||Create for Kylin 4.0.0-beta.|
If you are using Apache Kylin 4.0.1 and your $SPARK_HOME points to $KYLIN_HOME/spark, then you can ignore this document.
Kylin 4.0.1 will help you do the jar package replacement described in this document. You don't need to do these steps to start Kylin 4.0.1.
However, Kylin 4.0.1's automatic replacement of jar packages may fail.
If you encounter problems such as ClassNotFound during use Kylin 4.0.1, you still need to refer to this document to manually replace some jar packages.
Kylin on EMR 5.31
Create a EMR cluster
Check Hadoop version and download Kylin and Spark
Edit $KYLIN_HOME/conf/kylin.properties and add following content.
Prepare Metastore (Optional, only for test purpose)
Replace jars under $KYLIN_HOME/spark/jars
The Spark we downloaded is for Apache Hadoop 2.7, please replace with EMR-provided jars.
Modify $KYLIN_HOME/hadoop_conf/hive-site.xml (after Kylin instance started)
Kylin on EMR 5.31 with Working set to S3
EmrFileSystem not found
If you configure "kylin.env.hdfs-working-dir=s3://XXX/kylin/", you will faced ClassNotFoundException .
To fix this, please copy related jar from env.
- Monitor Pages
- Sparder(Query Engine) UI
Kylin on EMR 6.0.0
Each step is same other than the "Replace jars under $KYLIN_HOME/spark/jars" .