DSXchange

Thanks Craig.

Sajid,

Did you implement what we are suggesting?

Please let us know. It is working fine in my PX environment.

Still... regardless, for this technique to have any chance of working when you compare current and previous values, you need to set the 'previous' value last or they'll always be the same . ... Hi Craig, Please let me know if I am wrong. As I know all stage variables and derivation execute one afte...

Try this : if link.key <> stgOldKey THEN 1 ELSE stgCount + 1 ===> stgCount link.key ===> stgOldKey 1. Partition and order input link by key. 2. Keep the order when you defined the stage variable. 3. stgCount as Number and stgOldKey as Varchar/Number ( based on your input field type)

Try to connect oracle from your unix box using that connect string. If failed, talk to DBA otherwise talk to your DS admin.

Click on input tab of TD stage. And click on MultiLoad/TPump sub-tab.

Create two sequence one contain only Job and another contain your script.
Run them separately and let see.

APT_ORAUPSERT_COMMIT_ROW_INTERVAL

If you are not specifying SORT order it will effect the performance not the result. But if you not define partition it will effect the result.

Hi,
Could you please add a reject link to your transformer1 and try to compile. I read it in a forum ( don't know the proper reason), it resolved their issue of compilation. Just try it and let us know.

According to your output from command it is like "000000100.". Set SeqFile--->Format : Quote as none.

''''''

Try with bigint and let us know.

Add environment variable : APT_DISABLE_COMBINATION and set it value to true. Compile your job and run again.

When you stored seq number in the sequential file use data type as BigInt instead of decimal.

But if we are not specfying partition and sort in the input link, it all depends on these two environment variable : APT_NO_SORT_INSERTION
and APT_NO_PART_INSERTION. Means project dependent code. I think it is better to say "Input data must be partitioned and sorted" for join stage.