Installation Guide

Complete the following prerequisites before installing Datameer:

• Install the Hadoop client
  • Datameer application server, as well all data nodes, are configured properly with host names, DNS, datetime, NTP, and other details 
    
    • Datameer application server, as well all data nodes, have Java 1.8 (Oracle recommended)
      
      • Check this installation using the following commands: java -version and echo $JAVA_HOME
    • Datameer application server has Oracle Java Cryptography Extension (JCE) already installed. See Java SE Security for more information.
    • Commands such as hadoop, yarn, and mysql can be executed
• Install MySQL client
  • For Datameer's application database, the MySQL server must be prepared with necessary access 
• Grant administrative rights or root access
• Ensure Internet access to download packages and plug-ins or have necessary ZIP files downloaded and available
• If using Kerberos, configure a Kerberos Secured cluster for secure impersonation

Create the Datameer User

Administrative rights are required to create the Datameer user on the machine where Datameer is being installed. This can be accomplished under the root account. Make sure the user ID is above 500 and that the account has enough resources and file descriptors available. 

• Create the user account under which the Datameer service will be started and running later:

  Create user

  ./usr/sbin/groupadd --system datameer
  ./usr/sbin/useradd --system --create-home --gid datameer datameer

  These commands also create the directory /home/datameer. 

• Check the max number of open files - global level or per-user limits (or both) - and set it to 64K if it isn't set already. This configuration needs to be done on all nodes within the cluster and might require a reboot.

Create Directories for Application, Cache, Logs, and Temporary Files 

For performance reasons and to have better control about where space on the file systems and on disks is used, create separate directories for application, cache, logs, and temporary files. Do this according the Linux Filesystem Hierarchy Standard (FHS). To create the directories and change the permissions you need administrative rights. Complete this task under the user account root.

• Create the directories for application, cache, logs, and temporary files:

  Create directories

  mkdir -p /opt/datameer
  chown -R datameer:datameer /opt/datameer
  mkdir -p /var/cache/datameer
  chown -R datameer:datameer /var/cache/datameer
  mkdir -p /var/log/datameer
  chown -R datameer:datameer /var/log/datameer
  mkdir -p /tmp/datameer
  chown -R datameer:datameer /tmp/datameer

Switch the User and Change the Working Directory 

This should be the last task to which administrative rights are necessary. 

• Switch to the new Datameer user and change to the working directory where Datameer is being installed:

  Switch user and directory

  su - datameer
  cd /opt/datameer

  Proceed from within the Datameer installation directory and under the user account datameer only. 

Download and Unzip Datameer

Download the appropriate Datameer package for your Hadoop cluster distribution. If you have already a Datameer installation you can also start from here. 

• Download and unzip. 

  Switch user and directory

  curl -s -k -o Datameer-<package>.zip "https://download.datameer.com.s3.amazonaws.com/releases/Datameer-<version>/<dist>/Datameer-<package>.zip?<AWSproperties>" ; unzip Datameer* 

If you are an authorized Enterprise customer, you can get the download link for the latest public available package from https://my.datameer.com/workspace/downloads, or request one through your Customer Success Manager (CSM). 

Download and Install the MySQL Database JDBC Connector 

By default, the Datameer application runs with an HSQL file database that is created on the local filesystem under das-data/database/hsql-db. If you are setting up Datameer for production use, Datameer strongly recommends using MySQL instead of the HSQL file database. 
As of Datameer 7.4: MariaDB is supported as an alternative to MySQL.

• Download the official MySQL JDBC driver ZIP file, extract the driver from the archive file, and copy it into the correct destination:

  Download and install JDBC

  # Lookup latest JDBC driver version 
  JDBCDRV="$(curl -s -k 'https://dev.mysql.com/downloads/connector/j/' | grep -o -m 1 'mysql-connector-java.*zip')" 
  # Download latest JDBC driver version
  curl -s -L0 -k -O "https://dev.mysql.com/get/Downloads/Connector-J/${JDBCDRV}" 
  # Unzip driver package
  unzip mysql-connector* -d etc/custom-jars 
  # Move only the necessary JAR file 
  mv etc/custom-jars/mysql-connector*/*bin.jar etc/custom-jars
  # Clean up 
  rm -rf etc/custom-jars/mysql-connector-java-?.?.?? 

• Double-check if etc/custom-jars contain the latest mysql-connector-java-<version>-bin.jar:

  Check installation

  echo $JDBCDRV
  ll etc/custom-jars 

Configure Datameer for MySQL Database

Datameer service depends on the MySQL database. The MySQL database is used for writing to workbooks, permission changes, job execution, scheduling, and more. To function properly,  a response time should be between ten and twenty milliseconds. To run the application in MySQL mode, the following changes need to be implemented. As of Datameer 7.4: MariaDB is supported as an alternative to MySQL.

• Check database connection:

  Connection check

  mysqladmin version
  mysqladmin ping
  mysqladmin status
  echo q | telnet -e q `hostname` 3306
  nc -z -w1 `hostname` 3306

  You can follow up later with using the Check if the Datameer Application Database is Running and Accessible article.

• Initialize application database:

  Initialize database

  mysql -uroot -p < bin/mysql-init.sql
  mysql -uroot -p dap < bin/create-tables.sql

• Configure for an enterprise environment:

  etc/das-env.sh

  # Create a backup of the original configuration file 
  cp etc/das-env.sh etc/das-env.sh.original
  # Change the deploy mode 
  sed -i "s/\(DAS_DEPLOY_MODE=\).*\$/\1live/" etc/das-env.sh
  # Uncomment the database name you will be using 
  sed -i '/#.*DATAMEER_DB_NAME=/s/^#//' etc/das-env.sh 
  # Uncomment the user that the application should be started at 
  sed -i '/#.*DAS_USER=/s/^#//' etc/das-env.sh 
  # Specify the maximum size of the memory allocation pool 
  sed -i 's/Xmx2048m/Xmx4096m/' etc/das-env.sh
  # Create a log of changes made 
  diff -e etc/das-env.sh.original etc/das-env.sh > changes.das-env.sh

  If you aren't using sed, changes can also be made using other editors.
  Example: 
  Defining the DAS_USER is commented out in etc/das-env.sh. To use this property, uncomment the line and add for the correct value. 

Installing the License

If you have already received a Datameer license:

1. Launch the Datameer application and open the UI. The welcome page with all available licensing options is loaded. 
2. Press the button "Activate" and upload the key you received from Datameer. The license is being activated. You will be redirected to the login page. 

Start Datameer 

Start the Datameer service.

• Working within the current installation directory, use the following commands:

  Start Datameer

  # Start the Datameer service
  ./bin/conductor.sh start
  # Check the process ID (PID)
  ps -ef | grep -i "java.*jetty.*datameer" | grep -v grep | tr -s " " | cut -d " " -f2
  # Monitor the process booting and the log files
  cat logs/jvm-stdout.log; sleep 3; tail -F logs/`date +"%Y_%m_%d"`.stderrout.log logs/conductor.log

Stop Datameer 

Stop the Datameer service.

• Working within the current installation directory, use the following commands:

  Start Datameer

  # Stop the Datameer service
  ./bin/conductor.sh stop 
  # Monitor the process shutting down 
  cat logs/jvm-stdout.log; sleep 3; tail -F logs/`date +"%Y_%m_%d"`.stderrout.log logs/conductor.log

Restart Datameer

Restart the Datameer service.

• Working within the  current installation directory, use the following commands:

  Restart Datameer

  # Restart the Datameer service
  ./bin/conductor.sh restart
  # Monitor the process booting and the log files
  cat logs/jvm-stdout.log; sleep 3; tail -F logs/`date +"%Y_%m_%d"`.stderrout.log logs/conductor.log

Datameer Graceful Shutdown

Gracefully shut down the Datameer service.

1. Pause the Job Scheduler located under the Admin tab in Datameer.
2. Wait for current jobs to be marked as completed.
3. When all jobs have been completed, use the "stop" command on conductor.sh
4. After the Datameer application has been stopped, perform needed maintenance.
5. With all maintenance completed, resume Datameer using the "start" command on conductor.sh.
6. Under Datameer's Admin tab, resume the Job Scheduler. 

Service Check 

Check if the Datameer service is running and accessible. 

• Working within the current installation directory, use the following commands:

  Check service

  ./bin/conductor.sh check
  ps -ef | grep -i "java.*datameer" | grep -v grep
  lsof -i tcp@`hostname`:8080
  lsof -i tcp@`hostname`:8443
  lsof -i tcp | grep 'datameer'
  echo -e "GET /login \n\n" | openssl s_client -connect `hostname`:8443 -quiet | grep -i -m 1 'datameer'

• Monitor if the service is running and accessible later:

  Monitor service

  curl -k "https://`hostname`:8443/watchdog"
  lsof -i -p ps -ef | grep -i "java.*jetty.*datameer" | grep -v grep | tr -s " " | cut -d " " -f2

• You can also monitor the Datameer core directory size in HDFS.

Configure Datameer for Kerberos Secured Cluster

Before configuring Datameer for a Kerberos Secured cluster, test Kerberos authentication and job execution on CLI.  

• Send a test job to the cluster:

  Test job execution

  hadoop jar /<distribution-specific-path>/hadoop-mapreduce-examples-*.jar pi -Dmapreduce.job.queuename=root.default 3 10

To configure Datameer for a Kerberos-secured cluster follow the Secure Mode Configuration instructions. 

Secure Hadoop Distributed Filesystem (HDFS)

You must have a properly configured connection to a Kerberos-secured cluster to use the tool to secure the Hadoop Distributed Filesystem (HDFS) .

• For initial setup of secure impersonation, execute the following commands:

  Secure HDFS

  # Check current available access rights
  hadoop fs -ls /user/datameer
  # Configure Datameer Core Directories (aka Private Folder) 
  ./bin/secure_hdfs_tool.sh -u -g <dasuser>
  hadoop fs -chown -R datameer:<dasuser> /user/datameer/.staging 
  hadoop fs -chmod -R 770 /user/datameer/.staging
  # Check if changes are made correctly 
  hadoop fs -ls /user/datameer

Start Testing

Start the Datameer service to do final testing. 

• Working within the current installation directory, use the following commands:

  Start Datameer

  # Start the Datameer service
  ./bin/conductor.sh start
  # Check the process ID (PID)
  ps -ef | grep -i "java.*jetty.*datameer" | grep -v grep | tr -s " " | cut -d " " -f2
  # Monitor the process booting and the log files
  cat logs/jvm-stdout.log; sleep 3; tail -F logs/`date +"%Y_%m_%d"`.stderrout.log logs/conductor.log

Best Practices for Installing Datameer

Implement frequent database backups

Datameer service depends on the MySQL database, it is used for writing to workbooks, permission changes, job execution, scheduling, and more. It is highly recommend to backup the application database frequently.

0 * * * * mysqldump -u'dap' -p'dap' dap | gzip > /home/datameer/<company>_<system>_<datameer-version>_`date +\%Y\%m\%d_\%H\%M`.sql.gz

# Check from time to tome how long the database dump will take and if it fits into the timeslot
time mysqldump -u'dap' -p'dap' dap | gzip > /home/datameer/<company>_<system>_<datameer-version>_`date +\%Y\%m\%d_\%H\%M`.sql.gz
# Verify from time to time if the files are OK
gzip -d /home/datameer/company>_<system>_<datameer-version>_<date>_<time>.sql.gz 
head /home/datameer/<company>_<system>_<datameer-version>_<date>_<time>.sql

Enable and configure transport layer security (TLS)

Enable TLS for use with Datameer in production environments. As Datameer is packed with Jetty 9, you only need to enable modules.

• Enable TLS:

  Enable modules

  # Check default configuration
  java -jar start.jar --list-config | grep -i 'etc/jetty*'
  # Add SSL and HTTPS to the startup modules
  java -jar start.jar --add-to-start=ssl,https
  # Check final configuration 
  java -jar start.jar --list-config 

• Redirect all HTTP requests to HTTPS

• Configure TLS for Embedded Jetty (for more security)

• To change the HTTPS port follow the instructions under Configure TLS

  All port changes should be made in the start.ini file, which overrides jetty.port.

• Use your own custom certificate

Where to go from here

If you are a Datameer system administrator, see the Administrator's Guide.