DSXchange

Server routines can only be called in PX from BASIC transform stages, and there is a performance hit when you do that and is generally not recomended.

The routine CheckLine(Line,Delimiter,CheckCount) is

The "DS.TOOLS" entry in the VOC is a menu item that is installed in each DataStage project. If it is missing it is probably a sign that you have bigger problems. It is only installed in the project directories and not the master VOC; perhaps that is why you might not be able to find it. Also, even t...

Yes, setting the error code to 1 will do that, but won't put a message in the log stating why, just that the after-job triggered an abort.

In PX you would need to code this as a C++ routine. Since you can do the computation in a single line in a transform stage it would be more efficient and less work. Also, I think the DCOUNT() routine is a better one for you to use, i.e. a string of "Hello,World,!" would return 3, which is the number...

Read the as a single very big column to a transform stage.
Do a DCOUNT(In.BigCol,{YourColumnDelimiterCharacter}) to see if the number of columns is correct.
Using stage variables, pass the record on if the count is correct, otherwise store the string and prepend it to the next row's data.

Both DS and PX are based on different technologies. The DS Engine is an actual engine, based on the UniVerse product now owned by IBM. The PX Engine is different, as it isn't a server/engine as such but a collection of programs and files that are used to generate and run PX jobs. PX has been integra...

Use the Dataset tool option in your manager/director and view the schema of your dataset - it most likely was written with char() and your job metadata is integer. Correct one or the other.

Abu - that sounds a lot like hearsay. I've implemented jobs using IPC with huge numbers of records without problems. The IPC stage can improve performance on jobs that don't have inteprocess buffering enabled, if that is turned on then IPC stages make no difference at all.
Back the "Hang" question...

I dind't think that LONG was supported, but am not sure. How long is your definition?

You could write a short after-job routine, in which you get the row count from your database link and if the nubmer is below your threshold issue a call to DSLogFatal() to force the job to abort.

What are your first two column metadata definitions in your job? When you do a 'display data' do you only get the 1st column contents (correctly)? What is displayed for the 2nd column?

You need to be the Administrator to UNLOCK jobs. In addition, you cannot unlock just an INODE, you also need to specify the DEVICE. Furthermore, you should unlock the user, not the file.

I doubt there is a bug - this is base functionality used successfully in hundreds of sites.
Try pointing your output Seq file to '/dev/null'

You must do a "LOGTO {YourProjectName}", the DS.CHECKER verb is not present in the master VOC

I just realized you have the verb, but not the object code for it. You have a bad project, just as Ray has stated.

I really should have been paying attention - this is a Version 8 question and I think it likely that you are going to have to get IBM support involved as it does look more like a bug than a design issue.