DSXchange

A wise man gave the answer in this post...

viewtopic.php?t=142845

I guess it could be called a partitioning problem... remember that is you are running your job in on multiple nodes then you'll basically be performing the same calculation on each node... that is starting from 1 and adding one each time you'll encounter a new key. So, you'll need to include the par...

You need to do is define the start point and length in bracket after the source column and map it to your target column, for instance...

YourColumnName[1,4] mapped to year
YourColumnName[5,2] mapped to ...
YourColumnName[7,1]
YourColumnName[9,3]
YourColumnName[13,3]
YourColumnName[16,2]

Oracle source stage (SELECT xxx,yyy,zzz FROM ABC WHERE STAT='DONE' AND FLAG=0 AND TRUNC((sysDate-STRTDATE) / 365) >= 2) Lookup with reject link (on keys xxx,yyy,zzz) Oracle reference stage (SELECT xxx,yyy,zzz FROM DEF WHERE LEVEL <> 1) The MINUS result dataset would then be the output on the reject...

The year format for your input is not equal to the format %yyyy

Can't you problem be solved by setting the execution action to "Reset if required, then run" for all sub sequencers? Then it should only be necessary for support to reset the top sequencer...

Did your job finish with or without warnings?

Try to do a immidiate purge of the log. Run the job either from the Director or if you run it from the Designer then set the # of warnings limit to 50.

So... you only want to remove the double quote from "blank" fields?

Do you define a primary key for your table? If yes, then you have a unique contraint. But are you sure that you read ALL 140.000 records from the source succesfully? What is the record count from the source output link? Try to put a reject link on the sequential file stage and catch for catching any...

ray.wurlod wrote:You should be able to use "%yyyy%ddd" as the date format string. ...

Needless to say that this would be a much better solution

I presume it is defined as a string in the source? So, split the string 'YYYYint' into YYYY (where you change it to a date YYYY-01-01) and the int value Convert the date into a julian date Add the two int values together (I think you need to subtract 1) Convert the new julian date to the date format...

Well, you can use the modify stage for creating a duplicate column.
But my question is the same as Ronalds... Why not use the tranformer stage?

Read the connectivity guide for Oracle on mapping of oracle datatypes to datastage datatypes.

Try to look at your file in a HEX editor - maybe there is a character read with the last column that causes it to default in the transformer?

When did you run the job compared to the auto-purge settings in your project?