Find file
Fetching contributors…
Cannot retrieve contributors at this time
96 lines (58 sloc) 2.59 KB
How to compile FlumeBase
FlumeBase is a Java application with a maven-based build process.
BUILD SYSTEM REQUIREMENTS:
* Maven 2.0.9+
* Java 1.6+
* git
* Linux, or other POSIX-like environment (e.g. cygwin)
INSTALLING PREREQUISITES:
The following prerequisite tools must be installed prior to building
FlumeBase:
* Apache Thrift compiler 0.5.0
* maven-thrift-plugin 0.1.10
Installing Apache Thrift:
Apache Thrift can be downloaded from http://incubator.apache.org/thrift.
Download the 0.5.0 release of Thrift, untar it, and build it with
./configure && make && sudo make install.
Make sure that 'configure' indicates that Java libraries will be built.
You may need to download some of Thrift's build dependencies first. See
http://wiki.apache.org/thrift/ThriftRequirements (and the links at the
bottom of that page) for instructions on installing all Thrift build
requirements.
Correct installation can be verified with the following commands:
$ which thrift
/usr/local/bin/thrift
$ thrift -version
Thrift version 0.5.0
If you run 'thrift --help', you should verify that 'java'
is listed in the "Available generators" section.
Installing the maven-thrift-plugin:
To integrate thrift in the Maven build, a specific plugin must be
downloaded from github.com/flumebase/maven-thrift-plugin. Clone the
repository, build, and install the plugin by running the following:
git clone git://github.com/flumebase/maven-thrift-plugin.git
cd maven-thrift-plugin
mvn package install
RUNTIME PREREQUISITES:
FlumeBase requires Cloudera Flume 0.9.3 and Hadoop 0.20. These can
be installed from Cloudera's Distribution for Hadoop version 3b4 (CDH3B4).
Installation instructions for CDH are available at
http://archive.cloudera.com.
Installing from yum or apt-get is the most straightforward method. If
you install from release tar.gz files, then you will need to set
$HADOOP_HOME and $FLUME_HOME in your environment.
You can also run FlumeBase with a Cloudera Flume git repository. Clone
a copy of github.com/cloudera/flume, build it (run 'ant'), and set
FLUME_HOME to the path to your repository. You do not need to separately
install Hadoop in this case.
BUILDING FLUMEBASE:
Finally, build FlumeBase itself by changing to the FlumeBase source directory
(the directory containing this COMPILING.txt file) and running:
mvn package
Unit tests can be executed by running:
mvn test
(This is done as a sub-step of 'mvn package'.)
More thorough (longer-running) integration tests are available at:
mvn integration-test
The FlumeBase shell can be run with:
bin/flumebase shell