Making use of the Tez Binary Release tarball

  • If the binary tarball's name does not include anything referring to a hadoop version, then this implies that the tarball was compiled against the hadoop version that the Tez release compiles against by default. For example, for 0.7.0 and 0.8.0, the default hadoop version used is 2.6.0 ( this can be found by looking for the hadoop.version property in the top-level pom.xml in the source tarball for the release).

  • The tarball structure is as follows:

    apache-tez-{x.y.z}/
                      /tez*.jar
                      /lib/*.jar
                      /conf/tez*.xml.template
                      /share/tez.tar.gz
  • Set up Tez by following INSTALL.txt and use apache-tez-{x.y.z}/share/tez.tar.gz as the full tarball to be uploaded to HDFS.
  • Use the config templates under apache-tez-{x.y.z}/conf/ to create the tez-site.xml as needed in an appropriate conf directory. If you end up using apache-tez-{x.y.z}/conf/, then do an export TEZ_CONF_DIR="apache-tez-{x.y.z}/conf/"
  • Add "apache-tez-{x.y.z}/*:apache-tez-{x.y.z}/lib/*:${TEZ_CONF_DIR}" to HADOOP_CLASSPATH so as to get the tez client jars onto the classpath invoked when using the "bin/hadoop jar" command to run an example job.

 

  • No labels