IBM Mainframe Forum

Hi,

i have to DUMP DELETE datasets by using this sysin ADRDSSU



This batch is running once a month.

But this batch is during a very long time beacause of the lot's of entries in the catalogue af datasets :


more than 1 000 000 !

And each time, the batch have check all the datasets in the catalog to select them...

And there's only 1000 or 2000 datasets to dump delete each time.


My question is :

Do you know an other method to do it faster, or to prevent ADRDSSU to check each time the millions datasets.. ?

Thanks for your help.

If you've got a million datasets with the same three-qualifier prefix, me thinks there's something seriously wrong with your storage management.

Hello,

i know but i can't do anything.... It's like that...

please use a bit of logic...
how can <anybody/anything> filter on the creation date of a dataset without looking and checking at all of them ?
something to meditate about filtering is that not only the catalog is used, but also the VTOC is read...

evaluate if the <date cataloged> is a good alternative ...
in this case the VTOC access can b avoided, but....
You will have to build the include list Yourself using a REXX script to invoke IGGCSI00
and run the adrdssu job with the prebuilt list

I think the problem is not the amount of datasets, could you post the entire ADRDSSU's SYSIN?

Then I suggest that you

a) Think again
b) Read Enricos response and follow up on it

Click HERE to see how to use IGGCSIRX ( IGGCSI00 )

However, having played many times with this code, you will need to be quite proficient in REXX to accomplish this task

In fact, I have posted some REXX code which is readily available via google, just use IGGCSI00 and Expat as the search criteria

I think the problem is not the amount of datasets, could you post the entire ADRDSSU's SYSIN?

why do You think so? ... wiser to post evidence, not just wild guesses

by samb01's question:
Do you know an other method to do it faster, or to prevent ADRDSSU to check each time the millions datasets.. ?

I think the root reason of spend long time is DFdss logical DUMP command : WAIT.

If samb01's job not code WAIT, the default is WAIT(2,2), and my sugession is change it to WAIT(0,0).

So need he to post the complete SYSIN.

Alternate method:

Use Rexx and find out the dataset covering RACF profiles
Use LD P( Profile name ) gen all --> list out all the datasets.
put in the stem and use delete command to delete the datasets.

I am not sure how usefull it is ..just a suggession.....

Thanks,
Parthiban J