DSXchange

That whizzing sound is something again going over my head. At which release of DataStage are you allowed to create PX jobs on Windoze? I haven't seen this one yet, is it 7.5? Because, as far as I know, only connecting to a DS project on a Unix server with a PX license allows you to create or even i...

A hashed file stage input link can have the "create file" check box checked, in which case the Options button is enabled. Clicking this button opens the "Create file options" dialog. I would like to see an array of option buttons (32-bit and 64-bit) on this dialog box, with the default properly set ...

A Distributed file effectively gives you the UNION ALL of a number of hashed files. There are two prerequisites to being able to use a Distributed file: there must be no duplicate key values across the entire set of hashed files, and there must be some algorithmic method for detrmining in which part...

Arnd's advice only works in the BASIC Transformer stage. If you are using the regular parallel Transformer stage, the answer will depend on where you're reading the timestamp from. If it's a text file you can decompose it using substring techniques. If it's from a database you may also need to perfo...

If you develop server jobs you will run server jobs. Even on the UNIX machine, they will be server jobs. You can get some degree of parallelism by specifying your server jobs to have multi-instance capability; but in this case you need to manage partitioning of data yourself. There are only very lim...

Another possibility, since the record structure is identical (and only if the keys are disjoint), is to create a Distributed file as an "umbrella" over the four hashed files.

Try using IP addresses (xxx.xxx.xxx.xxx format) rather than host names. There may be an issue with name resolution in your network.

The short answers are: (1) you can change any of the values on the list (2) it's not documented anywhere (except in an advanced training class I wrote years ago but was canned by Informix USA under the "not invented here" rule) The first two options are about forcing the Aggregator stage to update i...

Not displaying the message of certain alerts is a known bug in DRS at versions 7.0 and 7.1 (at least - I haven't had 7.5 yet). Messages about the DataStage repository interface usually suggest that there is a problem in the repository, perhaps that its indexing needs re-building. Or that there is a ...

Using a UV stage and user-defined SQL you can create a UNION of the rows from all the hashed files. If there may be duplicate rows, specify UNION ALL. SELECT COL1,COL2,COL3 FROM HF1 UNION ALL SELECT COL1,COL2,COL3 FROM HF2 UNION ALL SELECT COL1,COL2,COL3 FROM HF3 UNION ALL SELECT COL1,COL2,COL3 FROM...

No. I meant exactly what I typed. @FM is to be preferred wherever possible. Particularly if you want to access elements somewhat randomly, because the hint structure is of most value in this case. The only time I would use @VM is where data are returned from a multi-valued field and I wanted to deco...

You CANNOT use the WITH trick for ordering data if it's an externally pathed hash file. It has to be a local account, so remove the fully qualified directory path and use a blank project/account name. It will then use the VOC pointer entry to find the path to the hash file. You can prove this very ...

Input rows are never rejected. Only output rows can be rejected. You can adapt your design to capture these by having a Transformer stage driving the output. Create an extra output link from this Transformer stage, and check the "Reject" box in the constraints grid for the new link. Ensure that this...

Is the output from the second Transformer stage going to a sequential file with "first line is column names" set? This could account for a "missing" row number 1.

Otherwise I'd have to try to replicate your design to comment further, and simply don't have the time at the moment.

To determine whether the DataStage deadlock daemon is running on a UNIX machine execute the UNIX command ps -ef | grep dsdlockd | grep -v grep If it's running you'll learn its process ID; if it's not running the report will be empty. Edit the $DSHOME/dsdlockd.config file to enable auto-start of the ...