Rate This Document
Findability
Accuracy
Completeness
Readability

Installing HiveExtension

The OmniOperator feature supports the Hive engine. You need to install the Hive engine on the cluster management node, and configure the HiveExtension dependency of openEuler on the management node and all compute nodes.

  1. Install the Hive engine. For details, see Environment Requirements.
  2. Download the HiveExtension plugin package and decompress it.

    Download boostkit-omniop-hive-3.1.0-1.5.0-aarch64.zip from Environment Requirements, decompress it to obtain boostkit-omniop-hive-3.1.0-1.5.0-aarch64.jar, and upload the JAR file to the management node.

  3. Install the HiveExtension dependency of openEuler.

    Configure the local yum source for each OS image and run the following commands to install the dependencies:

    yum install lz4-devel zstd-devel snappy-devel protobuf-c-devel protobuf-lite-devel boost-devel cyrus-sasl-devel jsoncpp-devel openssl-devel libatomic -y
  4. Configure HiveExtension.
    1. Obtain the ORC, Protobuf, Arrow, HDFS, and Parquet software packages from Environment Requirements and decompress them to obtain liborc.so, libprotobuf.so.24, libarrow.so.1100, libarrow_dataset.so.1100, libarrow_substrait.so.1100, libhdfs.so.0.0.0, and libparquet.so.1100. Obtain libLLVM-15.so and libjemalloc.so.2 from Installing Dependencies. Upload the preceding files to the /opt/omni-operator/lib directory on the management node and all compute nodes.
    2. Copy libboostkit-omniop-codegen-1.5.0-aarch64.so, libboostkit-omniop-vector-1.5.0-aarch64.so, libboostkit-omniop-operator-1.5.0-aarch64.so, and libsecurec.so to the /opt/omni-operator/lib directory on the management node and all compute nodes.
    3. Copy boostkit-omniop-hive-3.1.0-1.5.0-aarch64.jar to the /opt/omni-operator/lib directory and change the permission on the files in the directory to 550.
      chmod -R 550 /opt/omni-operator/lib/*
    4. Set the following temporary environment variables on the management node:
      export OMNI_CONF=/opt/omni-operator/hive
      export HIVE_AUX_JARS_PATH=/opt/omni-operator/lib
      export LD_LIBRARY_PATH=/opt/omni-operator/lib:$LD_LIBRARY_PATH
  • You can customize the /opt/omni-operator/hive directory of the OMNI_CONF variable. If SparkExtension is deployed in the environment as well, this directory must be different from the default /opt/omni-operator directory.
  • Each time a new shell session is started, you need to reset the environment variables in 4.