> I do not use any vendor's product., All my own set up, build and >configure.
My autobuild scripts should serve as readable documentation for this, since nearly everything's in a single Makefile with an install: target. Or take the easy route with $ make dist install In case you use the llap branch, just do "set hive.llap.execution.mode=none;" to use Tez. > java version "1.8.0_77" > Hadoop 2.6.0 ... > https://tez.apache.org/install.html Looks good so far. > Ok I just need to make it work as I have hive on spark engine as well. You're missing 3 things approximately - if you read through the Makefile in github. First, a good tez-site.xml in the classpath (remember, tez.lib.uris needs to be an HDFS path - for the rest, see the base file from autobuild). I usually update Tez to ${fs.default.name}/user/gopal/tez/tez.tar.gz and I do not use the minimal tarball, but the full dist tarball. The fixed tarball means it hits all the good localization characteristics of YARN, which can add up to minutes on a >250+ node cluster. Second, put that in the classpath for Hive (append to $INSTALL_ROOT/hive/bin/hive-config.sh) > export >HADOOP_CLASSPATH="$INSTALL_ROOT/tez/*:$INSTALL_ROOT/tez/lib/*:$INSTALL_ROO >T/tez/conf/:$HADOOP_CLASSPATH" > export HADOOP_USER_CLASSPATH_FIRST=true Replace $INSTALL_ROOT with wherever Tez is located. Third, disable the hive-1.x jars coming from SparkSQL (append/create in $INSTALL_ROOT/hive/conf/hive-env.sh) > export HIVE_SKIP_SPARK_ASSEMBLY=true After that, you can do > hive --hiveconf hive.execution.engine=tez to get Tez working (add --hiveconf tez.queue.name=<queue> to use queues). Cheers, Gopal
