I used Ambari from HDP to build a Hadoop cluster. On host1, it holds a NameNode and a DataNode. On host2, it holds another DataNode.
Then, outside this cluster, I installed a HDF in a host. I used NiFi to get data from remote SFTP and put data into remote HDFS. However, data can be only transferred to host1, the one with HDP. DataNode on host2 can't receive data. Also, there are no warnings and errors.
I have following questions.
1, I changed hadoop.proxyuser.root.hosts in hadoop configs. This helps me to successfully use Ambari File View UI to upload files into DataNode on host2. The user/groups of the uploaded file is admin/hdfs. Why? I understand that admin may refer to ambari admin, but why does modified root.hosts affect admin?
2, There is no user nifi in HDP. Why can I see nifi as owner of files in File View? How HDFS manage user and group?