Deletion of data in 2500 PS in single go

pankaj_kulkarni111 · New User Joined: 17 Aug 2006 Posts: 13 Location: Pune

Hi there

Is there any method to delete the data in 2500 PS files in a single go. I've a REXX which does this but is there any other method?

Cheers
Pankaj

Robert Sample · Posted: Wed Dec 24, 2008 2:27 am

What are you looking for -- a program to read the list & generate the deletes, one command to kill all 2500 files, or something else? If the first, which tools do you have at your site that you could use?

pankaj_kulkarni111 · New User Joined: 17 Aug 2006 Posts: 13 Location: Pune

Robert,

I'm looking for a program to read the list & generate the deletes in a single go.

Thanks
Pankaj

dick scherrer · Posted: Wed Dec 24, 2008 3:02 am

Hello,

Suggest you look into adrdssu/dfdss rather than rexx. . .

Anuj Dhawan · Posted: Wed Dec 24, 2008 9:49 am

Hi,

Technology won't limit you for such a task, little work around & it should be doable but what this requirement is all about - how those 2500 empty files will be used and why do you need them at first place ?

dick scherrer · Posted: Wed Dec 24, 2008 10:05 am

Hi Anuj,

Anuj Dhawan · Posted: Wed Dec 24, 2008 10:12 am

Hi Dick,

May be I misunderstood this then..

dick scherrer · Posted: Wed Dec 24, 2008 10:17 am

Hi Anuj,

I suspect the goal is to remove the datasets rather than just the data.

Maybe Pankaj will clarify. . .

expat · Posted: Wed Dec 24, 2008 1:12 pm

pankaj_kulkarni111 · New User Joined: 17 Aug 2006 Posts: 13 Location: Pune

Well thanks to you all first for your help!

The requirement is clear to me - There are 2500 datasets (PS and NOT PDS) and one of the project team need all them to be empty because they are going to store their data in it. Now as it's client I can't go back and ask them to do on their own, by emptying the dataset manually or via a program which deletes the earlier dataset and creates the new one. We are being paid for every technical work we are doing for them .

I don't want to delete those flat files but just the data in them - hope I'm clear here. I'm also trying to get this done, if I find any method I'll post it then.

Cheers,
Pankaj

enrico-sorichetti · Posted: Fri Dec 26, 2008 9:16 pm

Robert Sample · Posted: Fri Dec 26, 2008 9:26 pm

enrico: don't forget there's a limit on the number of steps in a single job. This will have to be done with multiple jobs (even if done with a program to open each file for output, the limit on DD statements per job would be reached).

enrico-sorichetti · Posted: Fri Dec 26, 2008 9:28 pm

Thank You Robert for reminding me
post edited to make a point of it

expat · Posted: Sat Dec 27, 2008 3:02 am

I would (and have done) use a REXX that dynamically allocates the output, then drops the allocation and keeps looping.

dick scherrer · Posted: Sun Dec 28, 2008 6:20 am

Hello,

I suppose there is something with this that completely escapes me, but i must admit that i'm confused on multiple levels. . .

If the files exist and are to be over-written as OLD files, why do anything to them at all? (oh, yeah, the client wants to pay for this) If there is some process that retains all of the dsns but not the space and the files were not originally allocated with sufficient secondaries to handle whatever the "new" data volume, there may be lots of x37's when data is written into them.

I don't recall a system that constantly "wrote over" existing ps files rather than deleting the old (if they exist) and allocating new with proper space allocations. What is to happen other than the first re-use of these 2500 files?

If some client wants to pay for a "thing", i suppose we should not discourage them (as i've been contracting for longer than most have even been in the field

). A sweet deal would be to have a fixed-price bid for "emptying" these 2500 at a price of $5 or $10.00 (USD) per file. Pretty good $ return for an hour or 2 of work. . .

And it may be that i am just hopelessly confused. . .

d

pankaj_kulkarni111 · New User Joined: 17 Aug 2006 Posts: 13 Location: Pune

Thanks for your help everybody, my problem is resolved by a REXX

I'll post that on Thursday, the new year day.

HAPPY NEW YEAR to you all

Cheers
Pankaj

expat · Posted: Mon Dec 29, 2008 3:08 pm

Dick, there were one or two times that I have had to do this, actually reset a file rather than delete and define because of some crazy special run scenario where the files were catalogued in two different catalogs on two different LPARs.

Some idiot set it up this way to do some testing, and then as the number of files grew, some poor storage guy

got stuck with nullifying the files for each test cycle.

REXX worked just fine for this.

dick scherrer · Posted: Mon Dec 29, 2008 8:21 pm

Hi Expat,

Might this not lead to a flock of x37s?

Might it not be both faster and less error prone downstream to backup a complete set of the "empty" files using adrdssu/dfdss or fdr and then simply restore them as needed?

I've needed to repeatedly "start fresh" in several situations, but forcing the file to empty hasn't (yet

) been part of the plan. . .

expat · Posted: Mon Dec 29, 2008 8:33 pm

Unfortunately not. The files were catalog'd in two different catalogs over two different LPARS, so if they were deleted/redefined on one LPAR then it would take ages to manually update the catalog on the other LPAR to keep all in sync.

It was a diabolical setup, and had they bothered to ask a techie in the first place we could have easily shared a catalog and the DASD.

Unfortunately the outlook at some sites is rather luddite and with the same people being in the same jobs in the same company for 20+ years they know no other approach. Still, it paid the bills for a good few months