You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@sqoop.apache.org by Sandish Kumar HN <sa...@gmail.com> on 2017/10/03 06:38:35 UTC
Re: Review Request 61882: SQOOP-3215 : sqoop import to hive table as
formats(avro, parquet)
> On Sept. 26, 2017, 1:35 p.m., Szabolcs Vasas wrote:
> > Hi Sandish,
> >
> > Thank you for your patch! Please find my findings below:
> >
> > - Sqoop uses Kite for importing in Parquet format so I think your patch should also utilize the Kite libraries for Hive table generation (see org.apache.sqoop.mapreduce.ParquetJob#createDataset). This would ensure that the table creation behaviour is the same when we import the data and when we only create the table.
> > - SQLServerHiveImportTest.testNormalHiveImportAsAvro test case fails (SQLServerHiveImportTest is a subclass of TestHiveImport so it has inherited your new test case). Can you please take a look?
> >
> > I have not had time yet to deeply review the Avro part but my impression is that it needs much more thorough testing(e.g. more data types) since it is a new feature. What do you think?
> >
> > Regards,
> > Szabolcs
- Do you want me to use org.apache.sqoop.mapreduce.ParquetJob#createDataset in testcases??
- Yes will add more data types on testcases for avro/parquet
> On Sept. 26, 2017, 1:35 p.m., Szabolcs Vasas wrote:
> > src/test/com/cloudera/sqoop/hive/TestHiveImport.java
> > Line 302 (original), 308 (patched)
> > <https://reviews.apache.org/r/61882/diff/1/?file=1802862#file1802862line308>
> >
> > Why did you need to add this change? As far as I understand your Parquet changes should only be effective if the --create-hive-table option is specified.
Yes I can remove this.
- Sandish Kumar
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/61882/#review186246
-----------------------------------------------------------
On Aug. 24, 2017, 9:51 a.m., Sandish Kumar HN wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/61882/
> -----------------------------------------------------------
>
> (Updated Aug. 24, 2017, 9:51 a.m.)
>
>
> Review request for Sqoop and Anna Szonyi.
>
>
> Bugs: SQOOP-3215
> https://issues.apache.org/jira/browse/SQOOP-3215
>
>
> Repository: sqoop-trunk
>
>
> Description
> -------
>
> sqoop import and create hive table as formats(avro,parquet)
>
>
> Diffs
> -----
>
> src/java/org/apache/sqoop/hive/TableDefWriter.java deec32d6
> src/java/org/apache/sqoop/tool/BaseSqoopTool.java 1564bdcb
> src/test/com/cloudera/sqoop/hive/TestHiveImport.java a624f52b
> testdata/hive/scripts/createhiveImportasparquet.q PRE-CREATION
> testdata/hive/scripts/normalHiveImportAvro.q PRE-CREATION
>
>
> Diff: https://reviews.apache.org/r/61882/diff/1/
>
>
> Testing
> -------
>
> testNormalHiveImportAsAvro, testNormalHiveImportAsParquet with table creation checks.
>
>
> Thanks,
>
> Sandish Kumar HN
>
>