without compression [numFiles=8, numRows=6547431, totalSize=66551787, rawDataSize=3154024078] with zlib [numFiles=8, numRows=6547431, totalSize=44046849, rawDataSize=3154024078] As you can see, the totalSize is less with zlib.
... View more
-rw-r--r--1 root root 2123 Sep214:31 ojdbc6.jar ojdbc.jar is much more than 2123 bytes. Most likely your download of the ojdbc.jar was not good
... View more
Make sure you add oracle jdbc driver to the sqoop sharelib in HDFS (find the current sharelib location using 'oozie admin -sharelibliblist sqoop' - For older installations (pre HDP -2.2) you can add the file to /user/oozie/share/lib/sqoop directory Make sure you either call oozie admin -sharelibupdate or restart oozie after that
... View more
You have a mismatch between yum cache and rpm db (most likely caused by interspersing yum and rpm commands). Do a yum clean all on the host and reinstall
... View more
Starting with HDP 2.2, hive-site.xml need not be added to the workflow as in earlier versions of Oozie as Oozie will automatically add the hive configuration if the action-conf/hive/ directory is properly setup. Ambari already handles this by copying the hive-site.xml to action-conf/hive/ directory whenever hive config changes are done. If you still want to copy your own hive-site.xml, add it to the job-xml instead of as a file as Oozie creates hive-site as part of launching the hive client
... View more
Starting with HDP 2.2, hive-site.xml need not be added to the workflow as in earlier versions of Oozie as Oozie will automatically add the hive configuration if the action-conf/hive/ directory is properly setup. Ambari already handles this by copying the hive-site.xml to action-conf/hive/ directory whenever hive config changes are done. If you still want to copy your own hive-site.xml, add it to the job-xml instead of as a file as Oozie creates hive-site as part of launching the hive client
... View more
Integrated authentication does not work with SQLServer even in a secure cluster with AD integration as the containers will not have the context (even in secure windows clusters, the impersonation level does not support this) You can use the password alias (which makes use of the Hadoop credentials support) to use SQL server authentication without exposing the passwords.
... View more
You can additional options to the --storage-stanza option. The storage stanza is just what gets appended to the create table statement and you can add syntactically valid options (like tblproperties)
... View more
If app is not a defined service for the TNS listener but the instance id, use this URL format jdbc:oracle:thin:@//<host>:1521:app Also, make sure you can connect to the Oracle instance from the all the client nodes (typically) as the sqoop mappers will run across the cluster.
... View more
