DSXchange

Dump the score - ascertain whether there are any inserted buffer operators. Without knowing that, it's impossible to advise. Larger buffers increase memory demand - smaller buffers decrease memory demand but also reduce throughput. Buffering is set (changed from default sizing) on the Advanced tab o...

APT_ParallelSortMergeOperator indicates a collector using the Sort Merge algorithm.

Use a hex display (for example od -x DATA.30 | head -1on UNIX) to examine the first four bytes of the DATA.30 file. This is the magic number. See above for how to decode.
adef010c indicates 32-bit, adef020c indicates 64-bit. Adjust for little-Endian byte order.

Read the full error message. It should tell you the name of the stage in which the error was generated.

It depends on many factors. Rather than try to canvas them all here, why not try what you have in mind, and report the results back to this thread?

Why bother with a DataStage job? All you need is an UPDATE statement in the database.

If I will do add another tranfromation my transfromation will take double time. That is not correct, and shows a weak understanding of how DataStage works. Without buffering, the Transformer stages will execute in the same process, and therefore add nothing to the overall run time - if you are perf...

dsenv is a shell script executed by all DataStage processes. It contains whatever you care to put in it. Typically it is used to set environment variable values, but I've also seen other commands in there, for example umask. .odbc.ini contains definitions of ODBC data sources. The above two files do...

Both my source data type and the target data type are numbers. Incorrect. The error message informs us that the input field is ustring[max=40]; that is, VarChar(40) with the extended properties set to "Unicode". The longest integer (int32) value is 10 digits long. DataStage warns you that there are...

DSguru2B wrote:Thats a nice piece of document Ray. Did you prepare it?

Yes, it's part of an EE Transformation Techniques module I've developed for a Parallel Job Techniques class/tutorial. The tutorial as a whole is not ready yet; I will announce its availability and price when it is.

And what does that suggest to you concerning the likely location of the problem (if not the cause)? The next step is to have your SAP gurus help you to identify any reasons (such as, but not limited to, locks or deadlocks or BAPI processes not running) that might cause the problem you encountered.

Use Fmt() in the server jobs to format the decimal numbers with as many leading and trailing zeroes as the parallel job expects. For example Fmt(MyNumber, "15'0'R2") specifies 12 digits to the left and two digits to the right of the decimal place.

The modify operator (and therefore the Modify stage) supports a string_trim() function.

A fairly complete list of Modify stage functions may be had here and then, of course, there are the manuals.

Here's a cute variation on the theme, retrieving the result as a single expression (for example to set a job parameter value). You need the job name from somewhere. Trans("DS_JOBOBJECTS", "J\" : Trans("DS_JOBS", JobName, 5, "X") : "\ROOT"...

You may load as many hashed files into the read cache in memory as you can fit in there. Once you exceed your limit, or if a hashed file is too large, you get a warning logged that the cache can not be used, and the hashed file will be accessed from disk. The size of the read cache can be tuned in t...