DSXchange

Add an after-job routine call that gets the row count from the TeraData selection using DSGetLinkInfo(DSJ.ME,{LinkName},DSJ.LINKROWCOUNT) and if it is 0 issue a call to DSLogFatal().

The server routine "KeyMgtGetNextValue" won't work in PX unless you revert to using a BASIC transform stage and that is not recommended. The surrogate key generation is not designed to store the highest value used after a run for use in the next one. The use of a database sequence is usually the mos...

Welcome to DSXChange!

If the files all share the same format (i.e. Copybook), then just combine all 4 files at the system level and read it in through the complex flat file stage if joining them is your goal.

Insert two dummy records into your Sybase table and re-run the job. Do you get 10,000, 10,001 or 10,002 records now?

and if you leave the output as Decimal(19,4) it doesn't return a null? I doubt it - doing any operation on a null value will return a null, so the assumption is athat you have a null in that column somewhere and the problem isn't being caused by a conversion or truncation error.

jingyu -please don't hijack threads, you already posted your problem on a separate one! Also, you should have added the extra information in your other post.

NLS is not required for the standard latin accent characters. Many times problems with NLS is assumed when the issue is only a display one. Where are you viewing this? There are many EBCDIC implementations and standards out there, but as far as I know they all convert the standard accents the same w...

How long does the same SQL select take from the DB2 command line? Has the table been modified recently and runstat'ed?

Yes, it is possible (perhaps even easy) to do something like this. The question is what sort of an alert message? You can put a transform stage in the stream that writes to this reject file, and add a specific call to write a log message (the sdk routine "UtilityWarningToLog" will do that). Or you c...

xinhuang66 wrote:varchar2(1000), is oracle column definition.

If that is the definition in your table, what is it in the job?
do you use the CIDN1 and CIDN2 columns somewhere in the job (i.e. do you have runtime column propagation issues)?

Umm, I'm still not sure what you mean. What is your definition in Oracle for this column?

I wasn't asking about the difference in the jobs, that was clear from your post. But you didn't state what the B job did differently from the A job. Could you explain "pulling data" in a bit more detail?

What is the Oracle DDL column length? Is the problem that you cannot load more than 300 characters into a varchar column or something else?

Please explain the differences in your job results.

Oops, I was thinking in terms of PX jobs, there are no peek stages in Server. Just make a copy of your job, replace the database writes with sequential file stages pointing to "/dev/null" and see if you still get a hang.