So Sqoop is trying to do things, that might seem surprising if you're new
to Hadoop: first, it's going to copy files containing the raw data to
/user/hive/warehouse/, and then second, it's going to execute a
CREATE TABLE (similar to what you may have used in SQL databases before)
with Hive to recreate the metadata that goes with those files.

In the output you showed me, it says the /user/hive/warehouse/
directory for the categories table already exists, and it's not expecting
it to. It seems that a previous run has failed for a different reason, and
we should clean it up before trying again.

To get rid of the raw data files, run: 'sudo -u hdfs hadoop fs -rm -r
/user/hive/warehouse/\*' (I'm assuming you don't have any other data in the
cluster you care about).

To get rid of the metadata, start the impala shell with 'impala-shell' (you
don't need any other arguments because you're on the Quickstart VM and the
defaults all happen to be correct). Run 'invalidate metadata;', and then
'show tables;'. For any tables you see, run 'drop table ;'.
Then rerun the Sqoop job and it *should* succeed, but if it doesn't the
output should give us the root cause of the real problem...