HDFS connection and HIVE connection for MapR 5.2


A SSH connection with the standard port 22 to a default MapR5 (MEP 3.0) installation is successful, e.g. hostname:22. It is no problem to upload and download files via the remote file nodes.

But I experience some problems to start a connection with HDFS Connector or HIVE Connector. I would expect a connection string like hostname:8020 or hostname:10000 to connect. The database name is e.g. dbname_db.db stored at /path/hive/warehouse/dbname_db.db .

I get errors like:

WARN HDFS Connection 0:83 Couldn't connect
ERROR Hive Connector 3:1 Execute failed: Could not create connection to database: Connection timed out: connect

Any idea?


Hello Frank,

MapR has its own implementation of HDFS which only supports connection via httpFS which is why you need to use the httpFS Connector node in KNIME instead of the HDFS Connector node. Regarding Hive you should be able to use the standard Hive Connector node with the build in driver and the standard settings. You only need to change the server ip and the user name.

Attached you can find an example workflow that can be used to connect to the MapR Sandbox 5.2.