DSXchange

We're with you but also clutching at straws. What has your official support provider had to say? Alas this is a scenario that would be difficult to reproduce - it would need comparable hardware but also the long time period over which to degrade the elapsed times. Have you been generating operationa...

Diagnosing errors from combined operators is only for the foolhardy. Disable operator combination (either at the job level or at the stage level) and run the job again, so that you can determine which stage (operator) is throwing the error. Then look for a VarChar data type that has no maximum lengt...

Most stages have a check box to enable RCP (runtime column propagation). Find it and clear it. Clearly RCP and "column propagation" are the same thing.

If not military time ("zulu"), then at least 24-hour clock, ideally using ISO 8601 standards (YYYY-MM-DD HH:MM:SS). You only get one format mask, but days and months can have one or two digits in them. That's why you need to force them to two digits by adding leading zeroes. Try to constru...

The devil's in the details, particularly if it's UNIX - Linux or vice versa. Little things, like different options for the find command in scripts, are what will bite you. The DataStage export/import should not give any problems at all, though you do need to create the projects (and set up their sec...

The UNIX command output has a line terminator, which will have been converted to a field mark. You need to trim this from the result before proceeding to use the result in an arithmetic expression.

Search the forums for how to accomplish this.

Hang on a minute. Doesn't the Merge stage require its inputs to be sorted? Then how come you need to sort again afterwards?

Do search for the particular error message, and/or look at the Options properties about generating warnings for unmatched masters.

Are you saying that you can add one stage type (Sequential File) without disturbing the link properties but cannot add a different stage type (Peek) without doing so? Do you mean, perhaps that you are replacing a database stage with a Sequential File stage? Then, at the very least, you are disturbin...

Not through a Mac, unless you've booted it in Windows. And even then I'm not sure - it's not something I've ever done. Too much hassle needing two hands to effect a "right button click" with a single-button mouse.

Why not?

The option is in the table definition. You create the table definition in isolation, where the option for multi-valued columns exists. Then you load this into the Hashed File stage. CREATE TABLE does create a hashed file. You don't need to do anything. For instructions on using ED consult the UniVer...

Maybe you just had horribly skewed data. Check with the Director's Monitor, with "Show Instances" enabled.

Any columns you don't KEEP or otherwise derive are dropped.

Any columns you don't DROP are kept.

The two are mutually exclusive, a point which is not well made in the Parallel Job Developer's Guide manual but is well made in the Orchestrate Operators manual.

No.

Is the extraction doing anything that might consume memory, such as sorting? Are there any nullable columns in the Oracle table that you are trying to load into non-nullable columns in the Data Set? Are you undertaking any transformation of the data as they flow through? Are you trying to load chara...