DSXchange

I see. You can always put a query such as I was looking for a way to do it as part of the read instead of as part of the processing.

If you enable interprocess buffering you can dispense with the IPC stage in this job. The only way that the sequential file can be written to and read from simultaneously is if you declare it as a named pipe (create it with mkfifo). Please check your sequential file setting on write and read to make...

What about your SQL SELECT clause, might you be usnig some conversion or CAST() in there?

You can set the attribute of columns (right-mouse click, then 'edit row') to explicitly set the default value of fields that are not interpreted correctly on input, as would happen with values of \000

Yes, hashed files can be copied - but they might not be useable afterwards. If you reference hashed files using a VOC entry the files cannot be copied unless you know how to change the VOC pointers. If you access the hashed file using an absolute path then you can copy it. A dynamic hashed file is a...

CHAR(4) perhaps?

I wonder if you might not have a non-displayable character at position 31? Have you tried doing a view-data and actually clicking on the field UNQ_VAL and seeing if you can move the cursor past position 30. Since you cannot VI a database, how are you generating the sequential file and the table inse...

This is your output representation for an invalid date format.

In this case you should not be using the FTP stage but executing the FTP command and subsequent rename directly. The method of putting a script around a FTP command is highly dependant upon your flavor of UNIX and I've found the best way to get a running script is to use Google to get a specific exa...

Could you post your actual derivation please?

I've encountered this limit on AIX before with very large projects and while the real solution is to split the project into multiple ones that are smaller and more manageable (As well as giving better response in the Director) you can resize some of DataStage's files to static hashed, this reduces t...

What have you defined the column as in the ODBC read stage.

Might you be reaching a 2Gb limit in the hashed file? Check the size of the DATA.30 and OVER.30 files in the directory. The only other cause might be that you have a @FM or @SQL.NULL in the key. Since you have ruled those out it is most likely the former.

PX has no line-level debugger as with Server jobs, but I usually make do by judicious use of PEEK stages to see the data. Which function did you use to convert from the original number/integer to VarChar as - I am surprised that you are getting leading spaces?

The default caching mechanism is 'private' so that message is just informational; any other type of cacheing would require configuration file changes and explicit enabling in the relevant jobs. Total size comes upto 1200MB and this fits into the 999MB memory. Ummm... how did you did you come to that...