How to update / clean DistributedCache when using Hue + Beeswax to run hive requests that define custom UDFs?

Question

I installed a Hadoop cluster (using the Cloudera distribution through the Cloudera Manager) and I run some Hive requests using the Hue interface that uses Beeswax at the bottom.

All my requests are running fine, and I even successfully deployed custom UDF.

But when deploying UDF, I came across a very disappointing version. In the initial version of my UDF class, I used a third-party class that raises a StackOverflowError .

I fixed this error and then confirmed that UDF can be deployed and successfully used from the command line.

Then, when I returned to using Hue and Beeswax again, I kept getting the same error. I could fix this only by changing the name of the Java UDF class. ( Lower to Lower2 ).

Now, my question is , what is the proper way to deal with such version issues?

From what I understand, when I add banks using the convenient form fields on the left, they are added to the distributed cache. So, how do I update / clear the distributed cache? (I could not get LIST JARS; etc. To run Hive / Beeswax from within. This gives me a syntax error.)

Answer 1

Since classes are loaded into the Beeswax Server JVM (the same thing happens with the JVM HiveServer1 and HiveServer2), deploying a new version of the banner can often require restarting these services to avoid such class loading problems.