11 Replies Latest reply on Jul 20, 2009 6:00 PM by 707883

[Jena] Debugging addInBulk()

707883 Jul 15, 2009 6:31 PM

Hello,

I am evaluating Oracle 11g with the Jena Adaptor 2.0. Currently I am trying to load customer records from one large file in N-TRIPLE format using addInBulk(). I did this successfully with a file with 100 sample records, and now I am trying it with 1000 (just a baby step towards ultimately being able to load all 200,000). Now I am getting the following error:

[java] - addInBulk: [171042 ] sqle
[java] java.sql.SQLException: ORA-01843: not a valid month
[java] ORA-06512: at "MDSYS.SDO_RDF_INTERNAL", line 781
[java] ORA-06512: at "MDSYS.SDO_RDF_INTERNAL", line 1886
[java] ORA-06512: at "MDSYS.SDO_RDF", line 294
[java] ORA-06512: at "MDSYS.RDF_APIS", line 542
[java] ORA-06512: at line 1
[java]
[java] at oracle.jdbc.driver.SQLStateMapping.newSQLException(SQLStateMapping.java:74)
[java] at oracle.jdbc.driver.DatabaseError.newSQLException(DatabaseError.java:131)
[java] at oracle.jdbc.driver.DatabaseError.throwSqlException(DatabaseError.java:204)
[java] at oracle.jdbc.driver.T4CTTIoer.processError(T4CTTIoer.java:455)
[java] at oracle.jdbc.driver.T4CTTIoer.processError(T4CTTIoer.java:413)
[java] at oracle.jdbc.driver.T4C8Oall.receive(T4C8Oall.java:1034)
[java] at oracle.jdbc.driver.T4CCallableStatement.doOall8(T4CCallableStatement.java:191)
[java] at oracle.jdbc.driver.T4CCallableStatement.executeForRows(T4CCallableStatement.java:950)
[java] at oracle.jdbc.driver.OracleStatement.doExecuteWithTimeout(OracleStatement.java:1222)
[java] at oracle.jdbc.driver.OraclePreparedStatement.executeInternal(OraclePreparedStatement.java:3387)
[java] at oracle.jdbc.driver.OraclePreparedStatement.execute(OraclePreparedStatement.java:3488)
[java] at oracle.jdbc.driver.OracleCallableStatement.execute(OracleCallableStatement.java:3840)
[java] at oracle.jdbc.driver.OraclePreparedStatementWrapper.execute(OraclePreparedStatementWrapper.java:1086)
[java] at oracle.spatial.rdf.client.jena.Oracle.executeCall(Oracle.java:474)
[java] at oracle.spatial.rdf.client.jena.OracleBulkUpdateHandler.addInBulk(OracleBulkUpdateHandler.java:685)
[java] at oracle.spatial.rdf.client.jena.OracleBulkUpdateHandler.addInBulk(OracleBulkUpdateHandler.java:472)
[java] at oracle.spatial.rdf.client.jena.OracleBulkUpdateHandler.addInBulk(OracleBulkUpdateHandler.java:439)
[java] at [redatced].OracleLoader.oracleBatchLoad(Unknown Source)
[java] at [redacted].OracleLoader.main(Unknown Source)
[java] java.sql.SQLException: ORA-01843: not a valid month
[java] ORA-06512: at "MDSYS.SDO_RDF_INTERNAL", line 781
[java] ORA-06512: at "MDSYS.SDO_RDF_INTERNAL", line 1886
[java] ORA-06512: at "MDSYS.SDO_RDF", line 294
[java] ORA-06512: at "MDSYS.RDF_APIS", line 542
[java] ORA-06512: at line 1

It's certainly possible that there is a glitch in my sample data - but my real question is, how do I debug and track down this problem? I am new to Oracle and I just can't figure out where to look and/or how to get more information about where in the data this is happening, what "month" it is choking on. Is there a log I could be looking at? Some way to turn on additional debugging output? Some clue as to exactly what statement is failing and what data is causing it to fail? I did some regex searching through the original N-TRIPLE source file to see if I could find something that looked like a date with a bad month, but no such date was apparent...

Thanks very much for any help!
Steve

1. Re: [Jena] Debugging addInBulk()

707883 Jul 16, 2009 12:54 AM (in response to 707883)

I have a bit more info - I added a StatusListener to the call to addInBulk(). This confirmed that all 171,042 triples were loaded (or so it claims) -- this error is happening in whatever post-processing happens after reporting the last triple loaded but before addInBulk() returns. This makes me suspect that this is actually a bug, or a configuration error on my part, rather than anything wrong with my test data.
Like Show 0 Likes(0)

Actions
2. Re: [Jena] Debugging addInBulk()

alwu-Oracle Jul 16, 2009 4:13 PM (in response to 707883)

Could you search in your data looking for dates and cut & paste them here?

Thanks,

Zhe Wu
Like Show 0 Likes(0)

Actions
3. Re: [Jena] Debugging addInBulk()

707883 Jul 16, 2009 5:48 PM (in response to alwu-Oracle)

I would, but even with the datatype URI's stripped off the file with only the dates (a text file where every line is something like "1941-03-21T00:00:00") is 1.4MB. Is there another way I could get that to you?
Like Show 0 Likes(0)

Actions
4. Re: [Jena] Debugging addInBulk()

alwu-Oracle Jul 16, 2009 6:43 PM (in response to 707883)

Assuming all the dates are of the same structure, could you build a simplified test case with just one triple containing one of those dates?
If it reproduces, then please cut & paste that triple.

Thanks,

Zhe Wu
Like Show 0 Likes(0)

Actions
5. Re: [Jena] Debugging addInBulk()

707883 Jul 17, 2009 1:19 AM (in response to alwu-Oracle)

Unfortunately I have not been able to reproduce this with any smaller dataset. I tried with a very small subset of this data (including a date triple) and it loaded fine. And again, I did successfully load about 17,000 triples' worth of similar data without incident.
Like Show 0 Likes(0)

Actions
6. Re: [Jena] Debugging addInBulk()

alwu-Oracle Jul 17, 2009 1:31 AM (in response to 707883)

I see. You can email this data set to alan dot wu at oracle dot com.

Cheers,

Zhe Wu
Like Show 0 Likes(0)

Actions
7. Re: [Jena] Debugging addInBulk()

Sdas-Oracle Jul 17, 2009 1:46 AM (in response to 707883)

Besides sending the data, if possible could you also try loading the data into a staging table using SQL*Loader and see if any of the triples get rejected and if not, then try bulk loading using the subprogram sem_apis.bulk_load_from_staging_table?
Like Show 0 Likes(0)

Actions
8. Re: [Jena] Debugging addInBulk()

alwu-Oracle Jul 20, 2009 4:26 PM (in response to 707883)

Hi,

If you have trouble sending the whole file, you can just send the third column of the N-Triple. That should be sufficient.

Cheers,

Zhe Wu
Like Show 0 Likes(0)

Actions
9. Re: [Jena] Debugging addInBulk()

707883 Jul 20, 2009 5:07 PM (in response to Sdas-Oracle)

I've been asked to hold off on my Oracle evaluation for now, so it may be a while before I have time to do any further debugging on my end. I have sent some data as Zhe Wu requested above; hopefully that will be useful.
Like Show 0 Likes(0)

Actions
10. Re: [Jena] Debugging addInBulk()

alwu-Oracle Jul 20, 2009 5:43 PM (in response to 707883)

Problem found. Actually, in your data, you have two lines contains something like "2008-002-01"
After removing those two entries, the addInBulk goes through without any problems.

Cheers,

Zhe Wu
Like Show 0 Likes(0)

Actions
11. Re: [Jena] Debugging addInBulk()

707883 Jul 20, 2009 6:00 PM (in response to alwu-Oracle)

ah, thanks for catching that. Sorry for bothering you with my bad data problem!
Like Show 0 Likes(0)

Actions

Go to original post