3.1. Building Sqoop2 from source code¶
This guide will show you how to build Sqoop2 from source code. Sqoop is using maven as build system. You you will need to use at least version 3.0 as older versions will not work correctly. All other dependencies will be downloaded by maven automatically. With exception of special JDBC drivers that are needed only for advanced integration tests.
3.1.1. Downloading source code¶
Sqoop project is using git as a revision control system hosted at Apache Software Foundation. You can clone entire repository using following command:
git clone https://git-wip-us.apache.org/repos/asf/sqoop.git sqoop2
Sqoop2 is currently developed in special branch sqoop2 that you need to check out after clone:
cd sqoop2 git checkout sqoop2
3.1.2. Building project¶
You can use usual maven targets like compile or package to build the project. Sqoop supports one major Hadoop revision at the moment - 2.x. As compiled code for one Hadoop major version can’t be used on another, you must compile Sqoop against appropriate Hadoop version.
Maven target package can be used to create Sqoop packages similar to the ones that are officially available for download. Sqoop will build only source tarball by default. You need to specify -Pbinary to build binary distribution.
mvn package -Pbinary
3.1.3. Running tests¶
Sqoop supports two different sets of tests. First smaller and much faster set is called unit tests and will be executed on maven target test. Second larger set of integration tests will be executed on maven target integration-test. Please note that integration tests might require manual steps for installing various JDBC drivers into your local maven cache.
Example for running unit tests:
Example for running integration tests:
For the unit tests, there are two helpful profiles: fast and slow. The fast unit tests do not start or use any services. The slow unit tests, may start services or use an external service (ie. MySQL).
mvn test -Pfast,hadoop200 mvn test -Pslow,hadoop200