So I am trying to import+overwrite a hive table with an incremental sqoop job. If I run it command line, as a regular user, everything works fine - command looks like this: sqoop import --connect jdbc:oracle:thin:@sanitized.hostname.com:1521:servicesb1 --username user --password-file hdfs://nameservice1/user/user/.password_bps --table SERVICESO.ACCT --compress --compression-codec org.apache.hadoop.io.compress.SnappyCodec --incremental lastmodified --check-column UPDATE_DATE --last-value "2016-02-10-22:08:18" --hive-import --hive-overwrite --hive-database service_inc --hive-table acct boom! overwrites the existing, drops the file. However, when I call it in oozie, it looks like it kinda sees hive, but not hive overwrite? Sqoop command arguments :
10975 [uber-SubtaskRunner] ERROR org.apache.sqoop.tool.ImportTool - Error during import: --merge-key or --append is required when using --incremental lastmodified and the output directory exists.
Intercepting System.exit(1) I have included hive-site.xml as a file in the oozie workflow, and set oozie.use.system.libpath=true and it doesn't error on the hive bit -- just that it doesn't accept the hive-overwrite flag. I tried ordering the arguments differently, but got the same error. I don't know where to look next, looking for help. Running Sqoop version: 1.4.6-cdh5.5.0, on CDH5.5 (obviously), no other strange things. I do see $SQOOP_CONF_DIR has not been set in the environment. in the log as well - but when i google/read about that I find (1) everyone says that's never the problem (2) I can't see how to set it in my workflow to make it go away in any docs or anything. Let me know
... View more
So, I did a dumb thing - I 'live' cloned one of our hosts to four other clones in VMware without thinking, and i didn't tell it to stop cloudera manager and it's associated services on startup after the clone. So, what I have is, 4 hosts with different host names, but all with the same UUID inside of the hosts - it will keep refreshing, point me to a different host when i click on it, and when i try to just 'remove' it. I remove host X, and Y still shows up, reports itself as X in the GUI, then as hostname Y, but when i remove from cluster/delete, it's confused again. I guess what i am asking, "Is there a way to delete these hosts on the 'backend' inside the CM database?" I.E. uninstall SCM, remove parcels, and delete it so i can 're-discover' it afterwards? I tried just reinstalling SCM, it then stopped on installing/activating parcels because the directories were already present.
... View more