MapR 5.0 Documentation : Install Sqoop2

Sqoop2 includes two packages:

  • the client package, mapr-sqoop2-client (install on each node that will act as a client)
  • the server package, mapr-sqoop2-server (install on at least one node in the cluster)

The Sqoop2 server also acts as a MapReduce client, so each node where you install the mapr-sqoop2-server package must also have Hadoop installed and configured. If you have a lot of client users, you can install multiple servers on multiple nodes, so as not to overload a single server.

Sqoop2 packages cannot be installed on the same nodes as Sqoop1 packages. However, you can use both versions in the same Hadoop cluster by installing Sqoop1 and Sqoop2 on different nodes.

Execute the following commands as root or using sudo:

  1. Update the list of available packages:

    On Ubuntu...
    apt-get update
    On RedHat and CentOS...
    yum clean all
  2. On each Sqoop2 server node, install mapr-sqoop2-server:

    On Ubuntu...
    apt-get install mapr-sqoop2-server
    On RedHat / CentOS...
    yum install mapr-sqoop2-server
  3. On each Sqoop2 client node, install mapr-sqoop2-client:

    On Ubuntu...
    apt-get install mapr-sqoop2-client
    On RedHat / CentOS...
    yum install mapr-sqoop2-client
  4. If you installed Sqoop2 1.99.3 (version prior to the 1507 release) and the cluster runs in YARN mode (MRv2), update the path to the Hadoop configuration directory in the file:


    The correct path for YARN is as follows:

    # Hadoop configuration directory

    By default the file contains the path for MRv1.

    As of Sqoop2 1.99.6-1507, Sqoop2 automatically sets the Hadoop configuration directory based on the Hadoop version that is used by the cluster. 
  5. Start each Sqoop server node.

    maprcli node services -name sqoop2 -action start -nodes <space delimited list of nodes>