Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Sqoop >> mail # user >> Does sqoop 1/2 support import as Parquet file?


Copy link to this message
-
Re: Does sqoop 1/2 support import as Parquet file?
To my understanding, there is not a "HCatalog" service in Cloudera Manager,
and I don't have to install hcatalog using RPM.

I tried HCatalog using sqoop1, but could not write Parquet format. Here are
what I did:

1. hadoop fs -mkdir /tmp/action_t
2. hive> create external table action_t ( ...) stored as parquet location
'/tmp/action_t';
3. sqoop import --connect jdbc:teradata://teraserver/DATABASE=PDMPUBLIC
org.apache.hadoop.io.compress.SnappyCodec -- --batch-size 1000

The problems are:
1. the job finished successfully, but the file in /tmp/action_t/_TEMP is in
text format.
2. If I use "--hcatalog-table action_text --create-hcatalog-table", the
data is not loaded into Hive.

Did I miss something?
On Tue, Jun 17, 2014 at 5:57 PM, Venkat Ranganathan <
[EMAIL PROTECTED]> wrote: