Upgrading Impala

You can upgrade from Impala 2.2.0 to Impala 2.5.0 on MapR version 5.1.0 or later. Version 2.5.0 of Impala for MapR introduces some new features.

Before you upgrade, verify that the system you upgrade on meets the prerequisites listed in the section Before You Begin and then complete the steps listed in the section Upgrading to Impala 2.5.0.

Before You Begin

Impala 2.5.0 for MapR has the following system requirements:

  • MapR version 5.1.0 or later
  • Hive 1.2
  • HBase 1.1.1
  • RHEL/CentOS 6.5, 6.6, 7.0, 7.1, 7.2 only (no Ubuntu, no SuSe)

If you want to upgrade to Impala 2.5.0 and you are currently running Impala 2.2.0 on MapR version 5.0.0 (or earlier) with Hive 0.13, you must complete the following steps in the order specified before you begin the upgrade process:

  1. Upgrade the MapR core package on all nodes to version 5.1.0 or later. For upgrade information, see Upgrading With the MapR Installer.

  2. Upgrade to Hive 1.2 and HBase 1.1.1. For upgrade information, see Upgrading Hive and Upgrading HBase.
  3. Verify that your system meets the required prerequisites. See Impala Prerequisites for a list of prerequisites.

If your system meets all of the requirements, continue to the next section, Upgrading to Impala 2.5.0.

Upgrading to Impala 2.5.0

To upgrade to the current version of Impala, complete the following steps:
Note: Always restart the Impala Statestore service prior to restarting the Impala server on nodes in the cluster.
  1. Stop the Impala Services.
    1. Run the following command to stop the Impala services on each Impala node:
      $ sudo maprcli node services -name impalaserver -action stop -nodes <node IP addresses separated by a space>
      Example:$ sudo maprcli node services -name impalaserver -action stop -nodes 10.10.30.166
    2. Run the following command to stop Impala the Statestore instances:
      $ sudo maprcli node services -name impalastore -action stop -nodes <node IP addresses separated by a space>
      Example:$ sudo maprcli node services -name impalastore -action stop -nodes 10.10.30.166
    3. Run the following command to stop the Impala Catalog instances:
      $ sudo maprcli node services -name impalacatalog -action stop -nodes <node IP addresses separated by a space>
      Example:$ sudo maprcli node services -name impalacatalog -action stop -nodes 10.10.30.166
  2. Upgrade Impala.
    1. Upgrade the mapr-impala package on all Impala nodes in the cluster:
      $ sudo yum upgrade mapr-impala
    2. In /opt/mapr/impala/impala-<version>/conf/env.sh, complete the following steps:
      1. Verify that the Statestore address is set to the address where you plan to run the Statestore service.
        IMPALA_STATE_STORE_HOST=<IP address hosting statestore>
      2. Change the Catalog service address to the address where you plan to run the Catalog service.
        CATALOG_SERVICE_HOST=<IP address hosting catalog service>

        Refer to Additional Impala Configuration Options for a list of other options that you can modify in env.sh.

    3. Verify that hive-site.xml has the following property configured on all nodes:
      <property>
          <name>hive.metastore.uris</name>
          <value>thrift://<metastore_server_host>:9083</value>
      </property>
      
    4. To upgrade the statestore service, issue the following command:
      $ sudo yum upgrade mapr-impala-statestore
    5. Install the catalog service using the following command:
      $ sudo yum upgrade mapr-impala-catalog
      Note: It is recommended (not required) that you install the catalog service on the same node as the statestore service.
    6. To upgrade the impala server, issue the following command:
      $ sudo yum upgrade mapr-impala-server
    7. Run configure.sh to refresh the node configuration.
      /opt/mapr/server/configure.sh -R
  3. Start the Impala services.
    1. Run the following command to start the Impala Statestore service on the designated host in the cluster:
      $ sudo maprcli node services -name impalastore -action start -nodes <node IP addresses separated by a space>
      Example:$ sudo maprcli node services -name impalastore -action start -nodes 10.10.30.166
    2. Run the following command to start the Catalog service on the designated host in the cluster:
      $ sudo maprcli node services -name impalacatalog -action start -nodes <node IP addresses separated by a space>
      Example:$ sudo maprcli node services -name impalacatalog -action start -nodes 10.10.30.166
    3. Issue the following command to start the Impala service on each Impala node:
      $ sudo maprcli node services -name impalaserver -action start -nodes <node IP addresses separated by a space>
      Example:$ sudo maprcli node services -name impalaserver -action start -nodes 10.10.30.166
    Note:

    Check the Impala log files for errors if the services do not start successfully. You can locate log files in the following Impala installation directory on each node:

    /opt/mapr/impala/impala-<version>/logs