Gluent Data Platform Daemons Installation

Introduction

Gluent Data Platform has two components that run as daemon processes:

1. Metadata Daemon (the metad process) is a daemon that collects information from Oracle, and Hadoop and Cloud backends to allow Smart Connector to construct an appropriate backend SQL statement

2. Data Daemon (the datad process) is a daemon that gives Smart Connector and Metadata Daemon the ability to communicate with Hadoop and Cloud backends

Metadata Daemon must run on all Oracle servers on which Gluent Data Platform is installed.

Data Daemon must run on at least one server on which Gluent Data Platform is installed. For production deployments it is recommended that Data Daemon does not run on an Oracle server.

Metadata Daemon Control

If the outcome of the Oracle Metadata Daemon OS User prerequisite check is that Metadata Daemon must run as root then:

1. Choose an appropriate method of running the daemon either as a service or using Oracle Clusterware (see Automatic Daemon Management)

2. Ensure METAD_AUTOSTART is set to false in offload.env

In environments where there is no requirement to run Metadata Daemon as the root user, no action is necessary. Smart Connector will check if Metadata Daemon is running and start it up when required.

Data Daemon

Note

The Gluent Data Platform images available on Google Cloud Platform Marketplace have Data Daemon pre-configured as a systemd service. If Data Daemon is running on a Gluent Node provisioned from a Google Cloud Platform Marketplace image then the service should be restarted to pick up the changes to the Gluent Data Platform Environment File. Refer to Control for the required start, stop and status commands.

Kerberos

If the backend authentication mechanism is Kerberos on either Cloudera Data Hub or Cloudera Data Platform Private Cloud, the following actions are required on all nodes where Data Daemon is to run.

Add the following entry to $OFFLOAD_HOME/conf/datad.conf (create the file if it does not exist):

JAVA_OPTS="-Djavax.security.auth.useSubjectCredsOnly=false"

Control

Data Daemon must always be started manually as there is no management of this daemon provided by Smart Connector or any other Gluent Data Platform process. A manual start can be achieved with:

$ $OFFLOAD_HOME/bin/datad &

If preferred, Data Daemon can be run either as a service or using Oracle Clusterware (see Automatic Daemon Management).

Automatic Daemon Management

Either of the Gluent Data Platform daemons can be managed automatically. Gluent document three approaches to achieve this:

1. Using Oracle Clusterware

2. Using Upstart

3. Using systemd

Before using any of these approaches the supplied configuration and script files must be updated.

Update Daemon Configuration Files

Important

Metadata Daemon actions must be performed on all Oracle servers where Gluent Data Platform has been installed. Data Daemon actions should be performed on all nodes where Data Daemon is to run.

Choose the appropriate command from the table below depending on which daemons will be running on the server on which the command is executed.

Each command will update the configuration and script files for the appropriate daemon(s) and change the ownership of those files to root:root.

The commands below must be run as root.

Gluent Daemon	Commands
Metadata Daemon	# export ORACLE_SID=<APPROPRIATE VALUE> # . <PATH TO offload.env> # cd $OFFLOAD_HOME/bin # ./connect --update-metad-files
Data Daemon	# . <PATH TO offload.env> # cd $OFFLOAD_HOME/bin # ./connect --update-datad-files
Both Daemons	# export ORACLE_SID=<APPROPRIATE VALUE> # . <PATH TO offload.env> # cd $OFFLOAD_HOME/bin # ./connect --update-root-files

Note

These connect commands only update files located in the Gluent Data Platform directory.

Oracle Clusterware for Managing Daemons

Gluent Data Platform includes scripts compatible with Oracle Clusterware Action Script for both Metadata Daemon and Data Daemon.

Installation

Note

These actions should be performed on a single Oracle RAC server only.

The following commands should be run as root.

Data Daemon
Add Resource 1	# crsctl add resource gluent-datad -type cluster_resource \ -attr "ACTION_SCRIPT=/u01/app/gluent/offload/scripts/datad/gluent-datad, DESCRIPTION='Data Daemon', START_DEPENDENCIES=hard(ora.orcl12.db), CARDINALITY=8"
Metadata Daemon
Add Resource 1	# crsctl add resource gluent-metad -type cluster_resource \ -attr "ACTION_SCRIPT=/u01/app/gluent/offload/scripts/metad/gluent-metad, DESCRIPTION='Metadata Daemon', START_DEPENDENCIES=hard(ora.orcl12.db), CARDINALITY=8"

1(1,2)

This command is to be run having sourced the appropriate environment for the Grid Infrastructure home. START_DEPENDENCIES, CARDINALITY, and ACTION_SCRIPT path should be set to appropriate values for the environment.

Control

Note

These actions should be performed on a single Oracle RAC server only.

The resource should now be started using the Start Resource command shown in the table below. Other control commands are shown for reference. All commands should be run as root.

Data Daemon
Start Resource	# crsctl start resource gluent-datad
Stop Resource	# crsctl stop resource gluent-datad
Status Resource	# crsctl status resource gluent-datad
Metadata Daemon
Start Resource	# crsctl start resource gluent-metad
Stop Resource	# crsctl stop resource gluent-metad
Status Resource	# crsctl status resource gluent-metad

Note

Please refer to Oracle’s documentation Making Applications Highly Available Using Oracle Clusterware for full details on using Oracle Clusterware to manage a third party application.

Upstart for Managing Daemons

Gluent Data Platform includes configuration files and scripts for both Metadata Daemon and Data Daemon for use with Upstart in Red Hat Enterprise Linux 6 based Linux distribution environments.

Installation

Note

These actions should be performed on all servers on which the Gluent Data Platform daemon is to run.

The following commands should be run as root.

Data Daemon
Copy configuration script	# cp $OFFLOAD_HOME/scripts/datad/gluent-datad.conf /etc/init
Metadata Daemon
Copy configuration script	# cp $OFFLOAD_HOME/scripts/metad/gluent-metad.conf /etc/init

Control

Note

These actions should be performed on all servers on which the Gluent Data Platform daemon is to run.

The resource should now be started using the Start Resource command shown in the table below. Other control commands are shown for reference. All commands should be run as root.

Data Daemon
Start Service	# start gluent-datad
Stop Service	# stop gluent-datad
Status Service	# status gluent-datad
Metadata Daemon
Start Service	# start gluent-metad
Stop Service	# stop gluent-metad
Status Service	# status gluent-metad

Tip

The presence of respawn in the configuration script indicates that Upstart will restart the Gluent Data Platform daemon if the process dies.

systemd for Managing Daemons

Gluent Data Platform includes configuration files and scripts for both Metadata Daemon and Data Daemon for use with systemd in Red Hat Enterprise Linux 7 based Linux distribution environments.

Installation

Note

These actions should be performed on all servers on which the Gluent Data Platform daemon is to run.

The following commands should be run as root.

Data Daemon
Copy configuration script	# cp $OFFLOAD_HOME/scripts/datad/gluent-datad.service /etc/systemd/system
Enable Service	# systemctl enable gluent-datad
Metadata Daemon
Copy configuration script	# cp $OFFLOAD_HOME/scripts/metad/gluent-metad.service /etc/systemd/system
Enable Service	# systemctl enable gluent-metad

Control

Note

These actions should be performed on all servers on which the Gluent Data Platform daemon is to run.

The resource should now be started using the Start Resource command shown in the table below. Other control commands are shown for reference. All commands should be run as root.

Data Daemon
Start Service	# systemctl start gluent-datad
Stop Service	# systemctl stop gluent-datad
Status Service	# systemctl status gluent-datad
Metadata Daemon
Start Service	# systemctl start gluent-metad
Stop Service	# systemctl stop gluent-metad
Status Service	# systemctl status gluent-metad

Tip

The presence of Restart=always in the configuration script indicates that systemd will restart the Gluent Data Platform Daemon if the process dies.

Documentation Feedback

Send feedback on this documentation to: feedback@gluent.com