RE: Removing All Spool Files For A User

midrange-l-request@xxxxxxxxxxxx wrote:

>   1. RE: Removing All Spool Files For A User (Dwayne Allison)
>
>date: Fri, 16 Sep 2005 07:34:30 -0500
>
>All of the IBM command would work, if the spool file was not so big.  The 
>system has to sort throughout the spool file before it can start deleting 
>them.  In the process of doing this we get a resource error.  So, we are 
>trying to create a program that will delete them one by one.  Instead of a 
>person deleting them one by one. 


Dwayne:

Been too busy to respond sooner...

I think Rob Berendt was the first (maybe only) one to mention that you might 
need to use the Open List of Spooled Files (QGYOLSPL) API rather than QUSLSPL.

That's because you're requesting a list that's larger than what can fit into a 
*USRSPC. You wrote "The system has to sort throughout..." and I can't tell if 
you meant that sorting was automatic or that you somehow wanted the list 
sorted. If you need the list sorted, you need to get all of the entries in one 
place and a *USRSPC won't do. You'll need to use QGYOLSPL to process the whole 
list and write the entries to something that can (1) hold all entries and (2) 
sort them. And maybe the sorting _is_ automatic based on your selection 
criteria.

Now, I think you _could_ use QUSLSPL if there is no need to sort and all you're 
going to do is DLTSPLF. Consider requesting the list with QUSLSPL and 
understand that the list would be 'P'artial. The header for the list should 
have a 'P' in the 'Information status' field. You'd run through the list and 
issue your DLTSPLF for each entry. When finished, you'd again call the API and 
repeat the process. This second call wouldn't have to list all that were just 
previously deleted. When you finally got a list that was no longer 'P'artial, 
you wouldn't call the API again; you'd just delete those last items and exit.

If you go ahead and use the more advanced Open List of Spooled Files (QGYOLSPL) 
API, I suppose it'd be easiest to write each entry to a physical file. You'd 
probably only write the fields that you needed from each entry. But because 
that makes each entry smaller, you might be able to fit the whole list into a 
*USRSPC yourself. Or you could create a *USRIDX that sorted in the order you 
wanted. Or...

Anyway, it seems reasonable that your problem is based in the limit on the 
number of spooled file entries that can fit in a *USRSPC. I think it's 
somewhere around 190K entries. (I've only bumped into it once.)

Tom Liotta