DSXchange

What values have you set for "Auto Write", "RCP", and "seperate"? If you are doing transfers explicitly then I guess they should all be "False".

It worked for me. Thats the reason i posted the version. The difference of 0.5 between DataStage 7.5 and 8.0 is making all the difference i guess.

1) Separating L from ETL basically means creating two jobs. One for Extraction and Transformation and writing the results to a DataSet. And a second job for Loading the data to the database from the DataSet. In ETL if the target database fails during the load then you will have to extract and transf...

ArndW wrote:

BugFree wrote:You can have a job with just the job control code...

No, you cannot have a PX job with just a sleep and compile it. ...

ArndW i have tried it. It works. But a warning message will be logged "main_program: Step has no operators". I have tried it. I am using 7.5.1A on HPUX.

The duplicates are because the stage is started on 2 nodes and each node will start from the 0 or 1 (not sure which) and continue. So if you run on 2 node you will get 2 duplicates and 4 node 4 duplicates.

You can have a job with just the job control code and no stages at all. You can actually run a job with SLEEP 10 in the job control and no stages on the canvas. It will wait for 10 seconds and finish successfully.

You can check if the length of the input string. If len(trim(In.Field)) = 19 Then StringToTimespamp(In.Field,"%yyyy-%mm-%dd %hh:%nn:%ss") Else StringToTimespamp(In.Field : " 00:00:00","%yyyy-%mm-%dd %hh:%nn:%ss") The above code will work only if the input is a timestamp...

The first warning says that the dataset.ds is not a data set descriptor. Where did you get this file from? Does the "View Data" work from inside the stage?
The second fatal error is probably due to the first one. Switching off RCP in the transformer will probably clear the last warning.

I don't know of any such tool. But "orchadmin dump" command can be used from the unix command line to view the data. There are options like -n to specify the number of records to be fetched from each partition, -part to specify which partition you want the records to be fetched. Do a "...

Ok. Here is the modified code. OpenSeq FilePath : FileName To TempFile On Error Ans = "Not Able to open file" Then Seek TempFile , 0, 2 THEN WriteSeq Message to TempFile Then Ans = "Success" End End CloseSeq TempFile Yuk? :oops: Ray I am still a beginner. Anyway thanks for the po...

This will work. OPENSEQ FilePath:FileName TO LogFile ON ERROR Ans = -1 THEN LOOP READSEQ DummyVar FROM LogFile THEN stat=STATUS() END ELSE EXIT END WHILE stat<>1 REPEAT WRITESEQ

You would have to use Status() function to go to the end of file. Look into OpenSeq(), ReadSeq(), and Status() functions.

You will have to use a combination of ReadSeq and STATUS() functions to loop through the file. STATUS returns 1 when it reaches the end of file. So that will be the condition for the loop.

1) In 'Insert only' mode is the rejects property set to true?
2) With 'Insert only' mode and rejects is there any entry in the log about 'bad' file? It is generally created in the scratch disk with the rejected records.

One row per file? I think it will be easier in command. Process the data in your job and write it to a file and then split the file using windows command.