RapidMiner

Hive Test Connection Exception

Highlighted
RMStaff

Re: Hive Test Connection Exception

Hi,

 

1) The "too many parallel JDBC connections are held" is an error that may appear after several unsuccessful connection attempts. It is not the original cause of the problem, but when it appears, you need to restart Studio, unfortunately, we hope to fix this very soon.

 

2) I guess the original problem is still the Hive connection timeout. First of all, I would increase the timeout values for the connection testing, as the quickstart VM that runs on the same machine may be slow. In Preferences, you can filter for "timeout", and increase these two Radoop settings, see the screenshot:

timeout.png

 

3) Check if quickstart.cloudera address can be accessed from the host machine, for example using ping or a web browser to access the Cloudera Manager (not from the guest, but the host machine).

 

4) You can check if the Hive connection request reaches the Hive service by checking the Hive service's Log on the Cloudera Manager interface (called Role Log). If you see log entries when RapidMiner runs the Hive connection test, then the request reaches the cluster, otherwise there may be a networking problem.

 

5) If you are using this cluster for RapidMiner (Radoop) only, you can safely remove several services that are not used by the RapidMiner. You can make the cluster much faster with that. To achieve that, you can Stop and Delete all services except HDFS, Hive, Spark, YARN (MR Included), ZooKeeper - these are needed by RapidMiner.

 

 

Best,

Peter