> On June 30, 2017, 1 p.m., Szabolcs Vasas wrote:
> > src/java/org/apache/sqoop/manager/DirectPostgresqlManager.java
> > Lines 281 (patched)
> > <https://reviews.apache.org/r/59710/diff/1/?file=1736285#file1736285line444>
> >
> >     The password is not set in the Configuration and because of this some 
> > test cases in PostgresqlImportTest fail.
> 
> Alex Ang wrote:
>     Added the password into the config.

I think adding the password is a bit trickier than just adding to the conf, 
PostgresqlImportTest still fails with the same error on my side.


> On June 30, 2017, 1 p.m., Szabolcs Vasas wrote:
> > src/java/org/apache/sqoop/manager/DirectPostgresqlManager.java
> > Lines 283 (patched)
> > <https://reviews.apache.org/r/59710/diff/1/?file=1736285#file1736285line446>
> >
> >     I am not sure why the JobConf object is needed here. The 
> > DBConfiguration constructor seems to expect a Configuration object, so conf 
> > could be passed as a parameter, couldn't it?
> 
> Alex Ang wrote:
>     DBConfiguration expects a org.apache.hadoop.mapred.JobConf and not 
> org.apache.hadoop.conf.Configuration. Without the wrapper, ClassCastException 
> will be thrown.

Are you sure you use the trunk branch? On trunk I can see this constructor: 
http://github.mtv.cloudera.com/CDH/sqoop/blob/cdh5-1.4.6/src/java/org/apache/sqoop/mapreduce/db/DBConfiguration.java#L266


> On June 30, 2017, 1 p.m., Szabolcs Vasas wrote:
> > src/java/org/apache/sqoop/manager/DirectPostgresqlManager.java
> > Line 505 (original), 312 (patched)
> > <https://reviews.apache.org/r/59710/diff/1/?file=1736285#file1736285line528>
> >
> >     I think the string encoding should be specified here to avoid using 
> > defaults.
> 
> Alex Ang wrote:
>     Could you advise where the encoding can be retrieved or I should default 
> it to UTF-8? Updated to UTF-8 in the new diff.

UTF-8 seems to be used in 
org.apache.sqoop.mapreduce.postgresql.PostgreSQLCopyExportMapper too but I 
think you should check what is the encoding used by the psql command because 
some users might rely on it.


> On June 30, 2017, 1 p.m., Szabolcs Vasas wrote:
> > src/java/org/apache/sqoop/manager/DirectPostgresqlManager.java
> > Line 506 (original), 313 (patched)
> > <https://reviews.apache.org/r/59710/diff/1/?file=1736285#file1736285line529>
> >
> >     Don't we need to apply the record delimiter here (see line 135 in the 
> > original file)?
> 
> Alex Ang wrote:
>     No. Applying the record delimiter will create additional row and fail the 
> unit test.

I see, but I am afraid if we do not use options.getOutputRecordDelim() then the 
change will not be backward compatible since some users might have specified a 
custom record delimiter here which seem to be ignored by your change.


- Szabolcs


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/59710/#review177994
-----------------------------------------------------------


On June 30, 2017, 5:17 p.m., Alex Ang wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/59710/
> -----------------------------------------------------------
> 
> (Updated June 30, 2017, 5:17 p.m.)
> 
> 
> Review request for Sqoop and Szabolcs Vasas.
> 
> 
> Bugs: SQOOP-3190
>     https://issues.apache.org/jira/browse/SQOOP-3190
> 
> 
> Repository: sqoop-trunk
> 
> 
> Description
> -------
> 
> Current Postgres direct import requires the installation of the psql utility 
> on all hadoop nodes. However, due to system constraints, this may not always 
> be possible. 
> It should be noted that Postgres provides a Java API which can execute the 
> same copy command as the psql utility. As such, it would be more ideal to 
> remove the psql utility dependency and switch to the Postgres Java API.
> 
> 
> Diffs
> -----
> 
>   src/java/org/apache/sqoop/manager/DirectPostgresqlManager.java 63b0704 
> 
> 
> Diff: https://reviews.apache.org/r/59710/diff/2/
> 
> 
> Testing
> -------
> 
> Tested with PostgresqlImportTest and passed all tests.
> 
> 
> Thanks,
> 
> Alex Ang
> 
>

Reply via email to