You should deploy the SAS In-Database deployment package
manually in the following instances:
-
Your Hadoop distribution is MapR.
-
Your Hadoop distribution is Cloudera
and any of the following is true:
-
Cloudera Manager is not installed.
-
You are not using Cloudera 5.8
or later.
-
Your other SAS software, such as
Base SAS and SAS/ACCESS Interface to Hadoop, was installed on Windows.
The SAS Deployment Manager cannot be used on a Windows client to install
the SAS In-Database deployment package.
-
Your Hadoop distribution is Hortonworks,
IBM BigInsights, or Pivotal HD, and any of the following are true:
-
Ambari is not installed or you
are using a version of Ambari prior to 2.4.
-
You are not using Hortonworks 2.5,
IBM BigInsights 4.2, or Pivotal 3.0 or later.
-
Your other SAS software, such as
Base SAS and SAS/ACCESS Interface to Hadoop, was installed on Windows.
The SAS Deployment Manager cannot be used on a Windows client to install
the SAS In-Database deployment package.
CAUTION:
Once you have chosen a deployment method, you should
continue to use that same deployment method when upgrading or redeploying
the SAS Embedded Process. Otherwise, the SAS Embedded Process can
become unusable.
For example, if you
use the SAS Deployment Manager to deploy the SAS software on the cluster,
you should continue to use the SAS Deployment Manager for upgrades
or redeployments. You should not use the manual deployment method
to upgrade or redeploy. If you do need to change deployment methods,
you must first uninstall the SAS software on the cluster using the
same method that you used to deploy it. You can then use the other
deployment method to install it.