1. Installation Guide
  2. Uninstalling Oracle Big Data SQL

6 Uninstalling Oracle Big Data SQL

If you want to completely uninstall Oracle Big Data SQL, the software must be removed from the Hadoop cluster management server and the Oracle Database nodes where it is currently installed.

If the Hadoop DataNode service is removed from a subset of the nodes, then Oracle Big Data SQL must be removed from those nodes only.

6.1 General Guidelines for Removing the Software

Oracle Big Data SQL can be uninstalled from the Hadoop cluster management server or from any Oracle Database servers connected to the cluster management server. The procedure is the same for all Hadoop platforms.

Guidelines for uninstalling Oracle Big Data SQL are as follows:

  • To perform a complete uninstall of Oracle Big Data SQL, remove the software from the cluster management server and from each Oracle Database server connected to the BDS service.

    This is a single script execution on each server. Not other manual steps are needed.

  • You can uninstall from the cluster management server first or from the database servers first.

    Tip:

    If you are removing the software from both sides of the installation, uninstall it from the database side first. If you uninstall from the cluster management server first, queries in process will fail.

  • On the database side, the first Oracle Big Data SQL connection installed is known as the “default cluster.”

Note:

In a Oracle RAC database, be sure to run the uninstall script on all nodes of the database.

6.2 Uninstalling From an Oracle Database Server

On any database server where you want to uninstall the current version of Oracle Big Data SQL, run the commands below as the database owner (usually the oracle user). In the case of a RAC database, be sure to perform the uninstall on each instance of the RAC.

If FUSE is enabled for TablesSpaces in HDFS, Uninstall it First

Note that the --crs command line parameter may or may not be required, depending upon conditions related to Grid support. See the Command Line Parameter Reference for bds-database-install.sh in this guide.

If you had previously set up FUSE to store Oracle tablespaces in HDFS, then you must first run the following command before uninstalling Oracle Big Data SQL:
$ ./bds-copy-tbs-to-hdfs.sh --uninstall
This command removes the FUSE mount point. You can then proceed with the Oracle Big Data SQL uninstall.

Uninstalling Oracle Big Data SQL

Run bds-database-install.sh with the uninstall command. 
$ /bds-database-install.sh --uninstall --crs=false

Note:

If Grid is not running on this database node, or, if the database does not use Grid (CRS/ASM), then include the optional --crs parameter and set it to false.

Also note that the --uninstall-as-primary and --uninstall-as-secondary parameters from previous releases are deprecated in this release. It is no longer necessary to differentiate between primary and secondary clusters in the uninstall process. In addition, --db-resource has been deprecated and is therefore not used in the example provided here.

6.3 Uninstalling From a Hadoop Cluster

Oracle Big Data SQL works with the Hadoop DataNode service. When you run the Oracle Big Data SQL installer on the configuration management server, it automatically propagates the Oracle Big Data SQL components to all Hadoop cluster nodes where the DataNode service is running. If the DataNode service is removed from one or more nodes, then you must also remove the Oracle Big Data SQL components from those same nodes.

  1. On the cluster management server (where CM or Ambari is running), log on as root.

  2. Change directories to the BDSjaguar directory under the location where the Oracle Big Data SQL Installation bundle BigDataSQL-<Hadoop distribution><version>.zip was extracted.

  3. Run Jaguar with the uninstall command and provide the configuration file previously used to install on the cluster (bds-config.json or other): 

    [root@myclusteradminserver:BDSjaguar] # ./jaguar uninstall bds-config.json
    When the uninstall is complete, the following message is returned:
    BigDataSQL: UNINSTALL workflow completed.