BDD uses a single script, called the orchestration script, to
install and deploy its components all at once. When the script finishes, BDD
will be completely installed and your cluster will be up and running.
The orchestration script is
contained in one of the BDD installation packages, which you will download to a
single directory on the Admin Server. You must perform the entire installation
process, including running the orchestration script, from this location.
The same installation package also contains the script's configuration
file,
bdd.conf, which defines the configuration of your
cluster and provides the script with information it requires at runtime. You
must update this file with information specific to your system and BDD cluster
configuration before you run the orchestration script.
Silent installation
Normally, when the orchestration script runs, it prompts you to enter
the following information:
- The username and password
for Cloudera Manager, which it uses to query Cloudera Manager for information
related to your CDH cluster.
- The username and password
for the WebLogic Server administrator. The script will create this user when it
deploys WebLogic.
- The username and password
for your Studio database, which it requires to connect Studio to the database.
You can avoid these steps by running the script in silent mode. To do
this, you must add the following environment variables to your system before
running the script. When the script runs, it checks for these environment
variables and executes silently if it finds them.
This table describes the environment variables required to run the
orchestration script in silent mode.
| Environment variable
|
Value
|
| CM_USER
|
The username for Cloudera Manager.
|
| CM_PASSWORD
|
The password for Cloudera Manager.
|
| WLS_USERNAME
|
The username for the WebLogic Server
administrator.
|
| WLS_PASSWORD
|
The password for the WebLogic Server
administrator. Remember that this must contain at least 8 characters, one of
which must be a number, and cannot start with a number.
|
| STUDIO_JDBC_USERNAME
|
The username for your Studio database.
|
| STUDIO_JDBC_PASSWORD
|
The password for your Studio database.
|
Orchestration script behavior
The following diagram illustrates the behavior of the orchestration
script.
Note: This diagram shows how the orchestration script distributes
various portions of the BDD installation packages on various nodes in the
deployment. This diagram is not intended to show how many nodes you can have in
your deployment. For various deployment scenarios, including options for
co-locating different parts of the BDD on the same nodes, see
Deployment configurations and diagrams.
When the script runs, it does the following:
- Reads and validates
bdd.conf.
- Prompts you for the user
names and passwords for Cloudera Manager, the WebLogic Server administrator,
and your database.
- Queries Cloudera Manager
for CDH-related information, including the hostnames and port numbers of
specific CDH nodes.
- Verifies that the Managed
Servers nodes and Dgraph nodes meet the minimum CPU and RAM requirements
defined in
bdd.conf.
- Verifies that the
COORDINATOR_INDEX defined in
bdd.conf does not exist.
- Verifies that the Hive
database defined in
bdd.conf exists.
- Distributes the
installation packages to each node in the cluster according to the
configuration defined in
bdd.conf.
- Verifies that each node
meets all other requirements, including the operating system, and the JDK.
- If the
FORCE property in
bdd.conf is set to
TRUE, deletes the
ORACLE_HOME directory from each node.
- Installs the components:
- Installs WebLogic
Server (including Studio and the Dgraph Gateway) on the Admin Server node and
all Managed Server nodes.
- Installs the Dgraph
and HDFS Agent on all nodes that will host Dgraph instances.
- Installs Data
Processing on the HDFS node and all Spark servers.
- Installs the Data
Processing CLI on all Managed Server nodes.
- Installs the
bdd-admin script on all Managed Server nodes,
Dgraph nodes, Spark worker nodes, and YARN node manager servers (not shown in
the diagram).
- Deploys Data Processing:
- Deploys Data
Processing to the HDFS node and all Spark nodes.
- Deploys the CLI to all
Managed Server nodes.
- If configured to do
so, deploys the Hive Table Detector to the specified node and starts it.
- Deploys WebLogic Server:
- Creates the WebLogic
domain and the Managed Servers.
- Deploys the Dgraph
Gateway and Studio as applications within the WebLogic domain.
- Deploys WebLogic as a
service on all Managed Servers.
- Starts all Managed
Servers.
- Deploys the Dgraph and
HDFS Agent:
- Deploys both
components as services to all Dgraph nodes.
- If configured to do
so, creates the empty Dgraph index files on the NFS.
- Starts the Dgraph and
HDFS Agent.
- Verifies that the entire
BDD deployment cluster is running.