DSXchange

Vijay, did IBM recommend VSS/PVCS for DataStage? If so, you should ask them exactly how they implemented it. These systems use "check in" and "check out" of objects (usually source files). Since DataStage uses a database to store jobs it is not directly possible to use these classical source control...

Krazykoolrohit wrote:eliminating DS should do the trick.

lol - those words are as close to heresy as you can get in this forum.

To the original question - can you query on the Teradata server itself to elminate the network from the equation?

You can use the SEQ() function in Server and PX transform stage to give you the numeric value of a character. Converting between ASCII and EBCDIC is done in Server with functions of the same name, but there is no equivalent builtin function to do that for you in PX. But the conversion is fairly stra...

There is no recovery, once you hit "delete" the job is completely removed. You will need to restore the job from a backup.

What kind of a selection/query are you doing? Can you duplicate the query in another tool of choice to see if the performance is as abyssmal?

Hello Rich, when I see ...the data is truncated, strange characters, zeros or blank... in this scenario I think of two main possibilities: 1. EBCDIC to ASCII conversion isn't being done anymore 2. NLS Settings might be different. Both might apply in your case, since the new server is on a Win2k box ...

splayer - when you use the graphical front-end to manage datasets it calls orchadmin; so you do have rights. From your shell, execute the $DSHOME/dsenv to set up your environment, run the $DSHOME/../PXEngine/bin/orchadmin ll {dataset}" command. The reason I asked you to do this is so that any error ...

DSGuru2B - good sed/awk solution. But the question remains open on how to do it in a CFF stage in PX. It should work, but I haven't gotten it to (yet).

You should capture your rejects, the procedure is documented both here and in your product documentation.

I might have jumped the gun, I assumed that the original poster meant that he/she had multiple engines (which use different ports) on the same machine. If that isn't the case I'd better remove my post as that would confuse matters.

What method in what stage are you using? Are you capturing rejects?

My first thought was yes, use a variable-length CFF definition and use something like the COBOL definition of: 01 DUMMY-REC. 12 RECLEN PIC 99 . 12 RECCON OCCURS 32 TIMES DEPENDING ON RECLEN PIC X. but then I tested in quickly on PX and am getting errors and currently don't have the time to look more...

use something along the lines of "dsjob -server :31540 -run -mode ..."

Does the same error happen when you export the whole project? If so, it might be due to one or more entities that you are exporting being corrupted.

splayer - is it faster if you use the UNIX command-line "orchadmin ll {dataset}" or is that also slow?