5 Expanding or Shrinking an Installation
Changes that occur to the Hadoop cluster over time, such as network reconfiguration or DataNode services added, moved, or removed, as well as changes on the Oracle Database server, can require corresponding changes to the Oracle Big Data SQL installation.
5.1 Adding or Removing Oracle Big Data SQL on Hadoop Cluster Nodes
On the Hadoop side, Oracle Big Data SQL must be installed on every cluster node where the DataNode role is running and should not be installed on other nodes. Note that when you run jaguar reconfigure
as described below, you must also reconfigure the Oracle Database side of the installation.
The Jaguar reconfigure
operaton uses the cluster management server to determine where the DataNode service is running and then installs or uninstalls Oracle Big Data SQL on each cluster node as needed.
Run reconfigure
when the distribution of DataNode services has changed in the Hadoop cluster.
# cd <Big Data SQL install directory>/BDSJaguar
# ./jaguar reconfigure <JSON_configuration_file>
BigDataSQL: RECONFIGURE workflow completed.
The following are situations where you should run reconfigure
:
-
New nodes are added to the Hadoop cluster and some or all of them are provisioned with the DataNode role.
-
Services and roles are redistributed in the cluster and some DataNode instances are moved.
-
One or more nodes provisioned with the DataNode role are removed from the cluster or deprovisioned.
Important:
Be sure to runreconfigure
after a cluster expansion. This will enable local processing for data stored on the updated nodes. Also, the new DataNode instances in the cluster provide additional resource for query processing that is not utilized until Oracle Big Data SQL is installed on those nodes.
You must also run bds-database-install.sh --reconfigure
on the database side after you run reconfigure
on the Hadoop side.
See Also:
The Jaguar Configuration Parameter and Command Reference provides a full description of the Jaguarreconfigure
operation.
Communicating DataNode Changes to Oracle Database
The Oracle Database side of the installation maintains an inventory of the DataNodes where Oracle Big Data SQL is installed in the Hadoop cluster. This is a map that tells it which nodes to query. If you run reconfigure
, this updated DataNode inventory must be provided to the database compute nodes. The reconfigure
procedure generates a new database installation bundle that includes the new inventory. Install this updated bundle on the database compute nodes using the same method that you use in a full installation. Install the update using bds-database-install.sh --reconfigure
-
As the database owner, copy the bundle to temporary location on the database node.
-
Extract the contents of the zip file.
-
Execute the run file extracted from the zip file.
-
Go to the installation directory under
$ORACLE_HOME/BDSJaguar-4.1.0.
-
Run the installer with the
--reconfigure
command line parameter. For example:$ cd $ORACLE_HOME/BDSJaguar-4.1.0/cdh510-6-node1.<domain>.com $ ./bds-database-install.sh --reconfigure
5.2 Adding or Removing Oracle Big Data SQL on Oracle Database Nodes
Add or remove Oracle Big Data SQL from Oracle Database nodes by manually running the install or uninstall program on individual database nodes as needed.
On the Oracle Database side, there is at this time no equivalent to the Hadoop-side Jaguar reconfigure
operation, which installs or uninstalls Oracle Big Data SQL on cluster nodes as required.
To Install Oracle Big Data SQL on Additional Nodes of Oracle Database RAC System
If you want to extend Oracle Big Data SQL from a Hadoop cluster to additional Oracle Database nodes at any time after the initial installation of Oracle Big Data SQL, then install the database-side installation bundle for that cluster on these database nodes using the same configuration applied to other nodes of the database.
Note:
The configuration for the connection between a Hadoop cluster and Oracle Database must be the same for all nodes of the database.-
You may find the bundle at <
Big Data SQL Install Directory>/BDSJaguar/db-bundles
or at the temporary location on other database nodes where you copied the bundle in order to stage the installation. The bundle ZIP file should include the Hadoop cluster name and a time stamp:bds-4.1.0-db-<cluster>-<yymmdd.hhmi>.zip
-
If you can no longer locate the original database-side installation bundle, but do have the configuration file that was used for other nodes of the database, then you can recreate the same bundle by running the Jaguar
reconfigure
command.[root@myclusteradminserver:BDSjaguar]# ./jaguar reconfigure my_original_configfile.json
Deploy and install the bundle on the new database nodes as described in Steps for Installing on Oracle Database Nodes.
To Uninstall Oracle Big Data SQL from Oracle Database Nodes