DSXchange

Epinzone,

I am unable to follow your requirement. Can you please explain your requirement in steps or pseudocode.

Also it will be useful to start a new thread for your own requirement.

Being v7.x, it must be your OS user.

Refer to admin guide

Craig,

You are right.

Tried a test job and gives me the exact error message.

But wonder what I did in Server jobs to avoid using all columns. Must try to remember.

Write your rejects into a single string and pivot it into multiple records.

Craig did beat me into the post.

Use round-robin partition method.

You will have multiple output links from the tx, each representing one possible error.

There are other methods to do this. But this will be simplest to design and implement - in my opinion.

Few basic questions

1.) What is the meta-data for the columns in DataStage?
2.) Are they extracted using same job or different ?
3.) Are you fetching directly from the table or from some view over it ?
4.) Are you viewing at the result after some transformation or view-data from Oracle stage ?

I think you can skip :1 if you are not using it.

But why do you have a comma before WHERE ?

one change in Ray's suggestion...

It will be value in specific count and zeros for all others.

For example
source1, flatfile, 10, 0, 0, 0
source2, db, 0, 10, 0, 0
...
....


tx -> agg (MAX of each columns)

send the reject rows from tx2 through to a funnel stage and its output to ds2

Your lines from log appear twice. Is it how it appears in the job log? You can do the following to analyse 1.) Try disabling the unique constraint and check the result after the run. 2.) Include a constraint to pass only one row from source and check the result 3.) Redirect the same output to sequen...

Can you provide the full transformation rule with the input and output values to explain the issue you are facing.

Without that others can do very little only.

Are you using user-defined SQLs ? That may be one of the reasons for translation. Pre-defined SQL tend to have To_Char() around the date values. Did you try mapping the values from flat-file to Oracle? DataStage automatically translates (it's internal) julian dates to Date of corresponding dbms. If ...

Search for IConv and OConv.

In case 2, input is date and not timestamp. So it fails.

In case 3, the output appears to be julian date.

Can you explain what you expect to see.