See the release notes page

The following install should take place on the API node of a v1.0+ IKANOW cluster, with the following Hadoop distribution installed:

• Any YARN based distribution (eg CDH5.x or HDP2.x) with the following services:
  • Storm, Zookeeper, Kafka, HDFS, MapReduce v2
  • Note the only distribution that ships all of the above out of the box is Hortonworks HDP 2.1+
    • For v2 only functionality, only a vanilla install is required
      • Don't forget to download the "site configuration" ZIP from Ambari and copy all the *-site.xml files into the local YARN config directory listed below (/opt/aleph2-home/yarn-config)
    • (for v1 analytics functionality, v1.0+ of the IKANOW platform is required and the following additional HDP install steps are required; otherwise just ensure that hadoop.standalone_mode=true in the v1 configuration, eg /opt/infinite-install/config/infinite.configuration.properties)

Download the RPM from the nightly builds or monthly releases and install (from the "Artefacts" tab)

• Copy all the files from the V1 Hadoop configuration directory into /opt/aleph2-home/yarn-config:

  • cp /opt/hadoop-infinite/mapreduce/hadoop/*.xml /opt/aleph2-home/yarn-config/
    • (If installing on an Infinit.e node running standalone Hadoop, then simply:
      • a) download the HDFS, YARN, MRv2 "site configuration" zips from Ambari/HDP, unzip, and copy the *site.xml files into `/opt/aleph2-home/yarn-config/``
        • (or you can get the XML files directly from /usr/hdp/current/hadoop-yarn-client/etc/hadoop/*-site.xml)
      • b) Run sed -i s/'${hdp.version}'/<HDP_VERSION>/g /opt/aleph2-home/yarn-config/*.xml
        • (Where "<HDP_VERSION>" can be obtained by doing hadoop fs -ls /hdp/apps/, eg "2.2.4.2-2"

• Copy defaults.yaml from the HDP storm configuration (eg from /usr/hdp/current/storm-client/conf/storm.yaml) into /opt/aleph2-home/yarn-config/storm.yaml (ie renaming it from defaults.yaml to storm.yaml)

• Copy zoo.cfg from the HDP zookeeper configuration (eg from /usr/hdp/current/zookeeper-client/conf/zoo.cfg)

• Configuring the distributed file system: using runuser hdfs -s /bin/sh -c "hadoop fs -mkdir -p <dir>", create the following directory structure:

  • /app
    • /app/aleph2
      • /app/aleph2/library
      • /app/aleph2/data

"Chown" /app/aleph2 recursively to tomcat (runuser hdfs -s /bin/sh -c "hadoop fs -chown -R tomcat /app/aleph2", using sudo if necessary)

• service ikanow-aleph2 start
• service ikanow-aleph2 stop
• service ikanow-aleph2 restart
• service ikanow-aleph2 status