Requirement When Installing VectorH on an HDFS Only Cluster
VectorH can be installed on a cluster that is running HDFS only--that is, running without other Apache Hadoop modules such as YARN, MapReduce, or Hive.
For VectorH to run on such a cluster, the Hadoop configuration must be modified.
The user who is the installation owner (default "actian") must belong to an operating system group that is named as the super user group for HDFS. Specifically, the Hadoop configuration file hdfs-site.xml must have the property dfs.permissions.superusergroup with the value “group” (you can choose any name), and in /etc/group, "group" must be defined and have the installation owner used as a member. For example:
hdfs-site.xml
<property>
<name>dfs.permissions.superusergroup</name>
<value>hdfsadmin</value>
</property>
/etc/group
hdfsadmin:x:1000:hdfs,actian
(The group number "1000" is arbitrary but cannot conflict with other group numbers.)
Last modified date: 01/26/2023