Edit the
master-install-location/gsInstaller/gsInstaller.properties
file and specify values for all of the following properties to install all HDP components:Note To perform minimal install (Apache Hadoop core components), specify values for Mandatory properties only (see third column in table).
Table 2.1. gsInstaller Configuration Properties Property Name Notes Example Mandatory/Optional/Conditional java64home Location of JAVA_HOME
for 64-bit JDK v 1.6 update 31 in your environment.Mandatory datanode_dir Comma-separated list of the DataNode’s data directories residing on separate disks. /hdp/1/hadoop/hdfs/data,/hdp/2/hadoop/hdfs/data Mandatory namenode_dir Comma-separated list of the NameNode’s data directories. Provide multiple directories on separate physical file systems and on separate mount points to preserve the NameNode metadata. /hdp/1/hadoop/hdfs/namenode,/hdp/2/hadoop/hdfs/namenode Mandatory snamenode_dir Comma-separated list of the NameNode’s checkpointing directories residing on separate disks. /hdp/1/hadoop/hdfs/snamenode,/hdp/2/hadoop/hdfs/snamenode Mandatory mapred_dir Comma-separated list of the MapReduce’s data directories residing on separate disks. /hdp/1/hadoop/mapred,/hdp/2/hadoop/mapred Mandatory log_dir Full path to Hadoop log directory. /var/log/hadoop
Mandatory pid_dir Full path to Hadoop PID directory. /var/run/hadoop
Mandatory sshkey Either provide full path to the sshkey which allows you to perform passwordless SSH OR Set this field to empty when passwordless SSH is set-up Mandatory installpig To install Pig, set the value to yes. (Default: yes) Optional installhbase To install HBase, set the value to yes. (Default: yes) Optional hbase_log_dir Location for HBase log directory. /var/log/hbase
Conditional. Required if installhbase
is set to yes.hbase_pid_dir Location for HBase PID directory. /var/run/hbase
zk_log_dir Location for ZooKeeper log directory. /var/log/zookeeper
Conditional. Required if installhbase
is set to yes.zk_pid_dir Location for ZooKeeper PID directory. /var/run/zookeeper
zk_data_dir Location for Zookeeper data directory. /hdp/1/hadoop/zookeeper
dbflavor Database flavor for Hive, Oozie, and Sqoop. Permissible values are mysql or oracle Conditional. Required only if installing either Hive, Oozie, or Sqoop dbhost FQDN for database host machine Conditional. Required only if installing either Hive, Oozie, or Sqoop jdbc_jar Location of the Oracle JDBC JAR file (see instructions here) Conditional. Required only if installing either Hive, Oozie, or Sqoop and if dbflavor
is set tooracle
dbsysuser User credentials for Oracle database system users dbsyspasswd installhive To install Hive, set the value to yes. (Default: no). You must first install and configure MySQL or Oracle for your cluster as instructed here. Also ensure that installhcat
property is set to yes.Optional. hive_log_dir Location for Hive log directory. /var/log/hive
Conditional. Required only if installhive
is set to yes.hive_dbname Database name for Hive metastore. Conditional. Required only if installhive
is set to yes.hive_dbuser Credentials for Hive Database user. hive_dbpasswd installhcat To install HCatalog, set the value to yes. (Default: yes) Optional. Ensure that installhive
property is set to yes.smoke_test_user Provide the value for the smoke test user. (Default: hdptestuser
)Mandatory installwebhcat To install WebHCat, set the value to yes. (Default: yes) Conditional. Ensure that installpig
,installhive
, andinstallhcat
properties are set toyes
.webhcat_log_dir Location for WebHCat log directory. /var/log/webhcat
Required only if installing WebHCat. webhcat_pid_dir Location for WebHCat PID directory. /var/run/webhcat
installsqoop To install Sqoop, set the value to yes. (Default: yes) Optional installoozie To install Oozie, set the value to yes. (Default: yes) Optional oozie_use_external_db Set this to yes, to use MySQL or Oracle database for Oozie metastore (default database is Derby). (Default: no) Conditional. Required only if installing Oozie oozie_log_dir Location for Oozie log directory. /var/log/oozie
oozie_pid_dir
Location for Oozie PID directory. /var/run/oozie
oozie_dbname Database name for Oozie metastore. oozie_dbuser Database user credentials for Oozie metastore. oozie_dbpasswd installFlume
To download Apache Flume RPM, set the value to yes. (Default: no). To deploy Flume, use the instructions available here. Optional installMahout
To install Apache Mahout set the value to yes.(Default: no) Optional enablemon Required if you want to install the HDP components for Monitoring (Ganglia and Nagios). (Default: yes) Optional. You must configure and deploy HDP Monitoring components as instructed in the next section.