DSXchange

The DSAddEnvVar() is not a function that can be called in DS/BASIC, it is a c program. I am not at a DS system right now and can't check what the equivalent BASIC program (if any) is. E

Hmm... if I got a 'file full' message while trying to write to a sequential file, I would look to see how much space I had on that device (particularly while the job is running, as the output file gets deleted by default upon job abort).

That is a good method, particularly if you have it working. You could export the project into a .dsx file and search that file in a text editor. Or write some code around DSGetProjectInfo(), DSGetJobInfo(), DSGetStageInfo() either in DS/Basic or in C++

What is the derivation of the stage variable "Run_Type_Parm" and can it be null?

please search on "Transaction Array Size" for a lot of posts on this Pubject. The command "dsjob" can be called from the Solaris (and HPUX,AIX,Linux,Windows) command line to start jobs.

My first guess is that you have not allocated sufficient table privileges for DB2 in your instance. PX requires the following: 2 Grant the DataStage users SELECT privileges on the system tables syscat.nodegroupdef, syscat.tablespaces, and syscat.tables. 3 Make the file db2nodes.cfg readable by the D...

I just went to www.ibm.com and did a search on "DataStage SOA" and got several references to red books, pdf files and other documents.

The first big hit is This page

That type of validation cannot be done in the key expression editor. It would best be added in a transform stage after the lookup.

Das, in order to store appreciably more than 2Gb in a hashed file you will, as you have already presumed, need to create a 64bit file. This cannot be done via the graphical front end, you will need to enter the administrator command line or the UV shell and issue a "CREATE.FILE {filename} DYNAMIC 64...

Are you using a 2-node configuration? Are you doing any re-partitioning?

If the modulus is too large the physical disk file wil become quite large, as space for the number of groups determined by the moduls has to be pre-allocated. Let us assume you have a file with 2 records that happen to hash to the first and last groups in the file. With a modulo of 2 your file would...

ICONV('THIS IS A BAD DATE','D4/E') is equal to 'THIS IS A BAD DATE'. Therefore just 1 ICONV() call is necessary to determine that you have an invalid date format, it is not necessary to issue 3 calls to do this.

The file "\\.\pipe\ora.4268.343000.fifo.0" is a named pipe {the key is the "fifo" i the name} which is used to send data from the DataStage processes. It does not contain any data and is deleted after job completion. I can remember trying to figure out where the .bad file was and cannot recall the l...

There is nothing that prevents you from looking up the same table twice. Please post in the correct forum (PX or Server) as they use lookups differently and also post what your "expected result" for the second lookup is and what you are actually getting.

It would seem that one (or more) of the SQL server columns are bigger than what you have defined in your job. Can you remove columns one-by-one to see which one it is complaining about and list the definition that the database has?