I'm facing a strange error I can't find in any logs on my Azure Hortonworks Sandbox. I'm trying to execute the following how-to to try the Mongodb integration with Hadoop (Hive) : https://www.mongodb.com/blog/post/using-mongodb-hadoop-spark-part-2-hive-example, but when I try to create a table in Mongo with Hive, I get the following error (using hive CLI, beeline and Ambari Hive view on both MR2 and Tez processor) :
Error: Error while processing statement: FAILED: Execution Error,
return code 1 from org.apache.hadoop.hive.ql.exec.DDLTask.com/mongodb/BasicDBObject (state=08S01,code=1)
Well, googling this message I found people quoting some java logs describing what the real exception is, and conclude (most of the time) the mongodb-java-driver jar were missing to the classpath.
My first question : Where are these logs describing the real exception? Because looking in /var/log/hive*/*.log and Yarn ResourceManagerUI -> logs, I can't see anything but "InterruptedException", "ERROR applicationhistoryservice.ApplicationHistoryServer (LogAdapter.java:error(69)) - RECEIVED SIGNAL 15: SIGTERM" and so on... But nothing really useful.
Next, I added the mongo-hadoop-core and mongo-hadoop-hive jars (built from github against my sandbox version specifications) AND the version 3.3.0 of mongodb-java-driver in a local path, specified in the section hive.aux.jars.path of the hive-site.xml, and finally restarted my hive service (so, including hive-server2)... But no way to get this working, and no way to get more info to understand why it's not running...
Am I forgetting something? Thanks for your help!