Not every port value that can be set in charts/hdfs-k8s/values.yaml is actually honored in the remainder of the chart files. For instance, hdfs-config-k8s sets dfs.datanode{,.http}.address only in the Kerberos setup, while in other scenarios DataNode ports are always set to their default values. Any variable that's ignored in this way is only useful for switching between Hadoop 2 and Hadoop 3 deployments, but any attempt to set custom values will fail. Another inconsistency: we are exposing ports only for certain services (e.g., NameNode), but not for others (e.g., DataNode).
Not every port value that can be set in
charts/hdfs-k8s/values.yamlis actually honored in the remainder of the chart files. For instance,hdfs-config-k8ssetsdfs.datanode{,.http}.addressonly in the Kerberos setup, while in other scenarios DataNode ports are always set to their default values. Any variable that's ignored in this way is only useful for switching between Hadoop 2 and Hadoop 3 deployments, but any attempt to set custom values will fail. Another inconsistency: we are exposing ports only for certain services (e.g., NameNode), but not for others (e.g., DataNode).