Read DSX to identify DB stages which write to hashed files

[clarcombe]

I have been looking at the DSX file in order to find out which jobs write to hashed files. This is part of a bigger process to move the hashed file creation to an overnight batch to reduce processing times.

In the section

Code: Select all

OLEType "CContainerView"

I can identify stage identifiers by

Code: Select all

StageList "V0S33|V51S0|...

and then their corresponding types by

Code: Select all

StageTypes "ID_PALETTEANNOTATION|ID_PALETTEJOBANNOTATION....

Then I can link the IDs to the Types and identify the Oracle stages by the CCustomStage tag and the hashed files by CHashedfileStage.

I then have to read through the DSX and find the hashed file

Code: Select all

Partner 

tag to see if it is linked to the

Code: Select all

CCustomstage

tag.

This seems all very longwinded (and not very reliable) and I wondered if there was a simpler way to do this ?

Any ideas ?

Thanks

[ArndW]

How about just searching the repository from the director for jobs that use the hashed file stage, or do you only want to locate writes?

In that case I would write a small BASIC program to traverse all jobs in a project looking for the specific stage - and I believe you can use the link information for the stage to determine if it is a source, reference or target stage.

[clarcombe]

When you say search Director, are you thinking of 8.x because I only have 7.X and no search capabilities.

Its not so much the HF stage but more ODBC-> HF or ORA->HF that I am concerned about. Given that there are several hundred jobs, I really need to identify just those types.

[ArndW]

My fault, I meant "Designer" not "Director" but my fingers disobeyed me.

From what you state, that search result won't help, you need to write BASIC job to

Code: Select all

FOR {all jobs in project}
   open job
   SELECT all passive stages
   FOR {all stages}
      IF Hashed file stage type 
      THEN
         SELECT all links for stage
         IF {no outgoing links, just input link}
         THEN
            PRINT 'We have a Winner!'
         END
      END
   NEXT stage

NEXT job

[clarcombe]

Without wanting you to write this, I don't know the syntax to interrogate the underlying Uniserv tables to get this information, hence my method of reading the DSX files.

1) Where can I get hold of this ?
2) Is it safe to interrogate the Uniserv tables this way ?

Thanks

[chulett]

Yes, it is perfectly safe to 'interrogate' (as in read) the legacy repository like that. Now, if you want to stick to the dsx method I would suggest something I used to great effect in the past. Chuck Smith has a variety of DataStage tools many of which parse a dsx export into something easily pushed into a relational target.

Specifically it was the "List all files and tables used by jobs in a dsx file" code that I modified to write to a database table so that it was accessible from SQL. From what I recall it should easily let you find jobs with target hashed files and database sources.

[clarcombe]

Excellent. Thanks Craig