Install the Analytics Environment

The SAS High-Performance Analytics environment components are installed with two shell scripts. Follow these steps to install:
  1. Make sure that you have reviewed all of the information contained in the section Preparing to Deploy the SAS High-Performance Analytics Environment.
  2. The software that is needed for the SAS High-Performance Analytics environment is available from within the SAS Software Depot that was created by the site depot administrator: depot-installation-location/standalone_installs/SAS_High-Performance_Node_Installation/3_6/Linux_for_x64.
  3. Copy the file that is appropriate for your operating system to the /tmp directory of the root node of the cluster:
    • Red Hat Linux (pre-version 6) and SUSE Linux 10:
      TKGrid_Linux_x86_64_rhel5.sh
    • Red Hat Linux 6 and other equivalent, kernel-level Linux systems:
      TKGrid_Linux_x86_64.sh
  4. Copy TKTGDat.sh to the /tmp directory of the root node of the cluster.
    Note: TKTGDat.sh contains the SAS linguistic binary files required to perform text analysis in SAS LASR Analytic Server with SAS Visual Analytics and to run PROC HPTMINE and HPTMSCORE with SAS Text Miner.
  5. Log on to the machine that is the root node of the cluster or the data appliance with a user account that has the necessary permissions.
  6. Change directories to the desired installation location, such as /opt.
    Record the location of where you installed the analytics environment, as other configuration programs will prompt you for this path later in the deployment process.
  7. Run the TKGrid shell script in this directory.
    The shell script creates the TKGrid subdirectory and places all files under that directory.
  8. Respond to the prompts from the shell script:
    Configuration Parameters for the TKGrid Shell Script
    Parameter
    Description
    TKGrid Configuration Utility.
    Running on 'machine-name' Using stdin for options.
    Shared install or replicate to each node? (Y=SHARED/n=replicated)
    If you are installing to a local drive on each node, then specify n and press Enter to indicate that this is a replicated installation. If you are installing to a drive that is shared across all the nodes (for example, NFS), then specify y and press Enter.
    Enter additional paths to include in LD_LIBRARY_PATH, separated by colons (:)
    If you have any external library paths that you want to be accessible to the SAS High-Performance Analytics environment, enter the paths here and press Enter. Otherwise, press Enter.
    Enter NFS mount to MAPR directory (ie: /mapr/my.cluster.com, default is none).
    If you want the analytics environment to be able to read and write MapR data directly, enter the NFS mount here (for example, /mapr/my.cluster.com).
    The mount point must exist on all nodes, including the name node.
    The TKGrid script sets the environment variable, TKMPI_MAPRHDFSPREFIX, to point to this share.
    Enter additional options to mpirun.
    If you have any mpirun options to add, specify them and press Enter.
    If you are using Kerberos, specify the following option and press Enter:
    -genvlist `env | sed -e s/=.*/,/ | sed /KRB5CCNAME/d | tr -d '\n'`TKPATH,LD_LIBRARY_PATH
    Note: Enter the above option on one line. Do not add any carriage returns or other whitespace characters.
    If you have no additional options, press Enter.
    Enter path to use for Utility files. (default is /tmp).
    SAS High-Performance Analytics applications might write scratch files. By default, these files are created in the /tmp directory. To accept the default, press Enter. Or, to redirect the files to a different location, specify the path and press Enter.
    Note: If the directory that you specified does not exist, you must create it manually.
    Enter path to Hadoop. (default is Hadoop not installed).
    If your site uses Hadoop, enter the installation directory (the value of the variable, HADOOP_HOME) and press Enter. If your site does not use Hadoop, press Enter.
    Force Root Rank to run on headnode? (y/N)
    If the appliance resides behind a firewall and only the root node can connect back to the client machines, specify y and press Enter. Otherwise, specify n and press Enter.
    Enter full path to machine list. The head node 'head-node-machine-name' should be listed first.
    Specify the name of the file that you created in the section List the Machines in the Cluster or Appliance (for example, /etc/gridhosts) and press Enter.
    Enter maximum runtime for grid jobs (in seconds). Default 7200 (2 hours).
    If a SAS High-Performance Analytics application executes for more than the maximum allowable run time, it is automatically terminated. You can adjust that run-time limit here.
    To accept the default, press Enter. Or, specify a different maximum run time (in seconds) and press Enter.
    Enter value for UMASK. (default is unset.)
    To set no umask value, press Enter. Or, specify a umask value and press Enter.
    For more information, see Consider Umask Settings.
  9. If you selected a replicated installation at the first prompt, you are now prompted to choose the technique for distributing the contents to the appliance nodes: 
    The install can now copy this directory to all the machines
listed in 'filename' using scp, skipping the first entry.
Perform copy? 
(YES/no)
    Press Enter if you want the installation program to perform the replication. Enter no if you are distributing the contents of the installation directory by some other technique.
  10. Next, in the same directory from which you ran the TKGrid shell script, run TKTGDat.sh.
    The shell script creates the TKTGDat subdirectory and places all files in that directory.
  11. Respond to the prompts from the shell script:
    Configuration Prompts for the TKTGDat Shell Script
    TKTG Configuration Utility.
    Running on 'machine-name'
    Using stdin for options.
    Shared install or replicate to each node? (Y=SHARED/n=replicated)
    If you are installing to a local drive on each node, then specify n and press Enter to indicate that this is a replicated installation. If you are installing to a drive that is shared across all the nodes (for example, NFS), then specify y and press Enter.
    Enter full path to machine list.
    Specify the name of the file that you created in the section List the Machines in the Cluster or Appliance (for example, /etc/gridhosts) and press Enter.
  12. If you selected a replicated installation at the first prompt, you are now prompted to choose the technique for distributing the contents to the appliance nodes: 
    The install can now copy this directory to all the machines
listed in 'filename' using scp, skipping the first entry.
Perform copy? (YES/no)
    If you want the installation program to perform the replication, specify yes and press Enter. If you are distributing the contents of the installation directory by some other technique, specify no and press Enter.
  13. If you are planning to use the High-Performance Analytics environment in a locale other than English, you must copy the appropriate locale files from SASFoundation/9.4/misc/tktg to the TKTGDat directory on every machine in the analytics cluster.
    In this example, the simultaneous command, simcp, is used to copy the Japanese locale files to the TKTGDat directory on each machine in the analytics cluster: 
    /opt/TKGrid/bin/simcp /opt/SASHome/SASFoundation/9.4/misc/tktg/jp* /opt/TKTGDat
  14. Make one of the following choices:
Copyright © SAS Institute Inc. All Rights Reserved.
Last updated: June 19, 2017