On Fri, 20 Nov 2009 21:25:38 -0700, Sam Siegel wrote:
>This is exactly what I've recently done and it works perfect on 1.09
>
How did you verify that the data set was not recalled before it
was deleted.
>On Fri, Nov 20, 2009 at 6:29 PM, Donald Russell wrote:
>
>> //S1 EXEC PGM=IEFBR14
>> //FILE1 DD DSN=...,DISP=(OLD,DELETE),UNIT=(,,DEFER)
>>
>> I believe this will delete a migrated dataset without recalling it, because
>> IIRC the DEFER option on UNIT says, don't allocate a device for this until
>> the dataset is opened... since IEFBR14 doesn't open the dataset, it's not
>> allocated to a device and there's no need to recall it.
>>
>> YMMV... It's been a while. :-)
>>
This is partly rendered moot by changes in 1.11.
I suspect this doesn't work as you wish for a couple reasons:
o Verifiably, on 1.10. I migrated a data set, then ran the job step:
//STEP EXEC PGM=IEFBR14
//D DD DISP=OLD,UNIT=(,,DEFER),
// DSN=&SYSUID...
The data set was recalled. Even a better test: We have a test
system on which HSM is not enabled. There, I submitted the
job with DISP=(OLD,DELETE),UNIT=(,,DEFER). There, it hung on WTOR:
ARC0050A DFSMSHSM IS NOT ACTIVE - START DFSMSHSM
o Suspected; not verifiable. Delete processing in allocation may
require that the data set reside on a real volume, not merely be
defined in the MCDS.
The extent to which this is not moot is that some parsimonious
programmers choose not to add an IEFBR14 step, but to add the DD
statement with DISP=(OLD,DELETE) to some other job step. I've
done that at times.
It might have been better had z/OS design chosen to honor the
DEFER specification for DASD data sets, consistent with the
behavior for tape, leaving the programmer with control of whether
or not the data set is recalled at step allocation. This would
have supported shortcut deletion on a non-IEFBR14 step and
alleviated the phobia about misbehavior of a bogus IEFBR14.
(Doesn't the IBM-controlled prefix, "IEF" caution sites against
substituting an "IEFBR14" with unconventional behavior?)
Hmmm. If a form of DEFER were tolerated on a Unix PATH= specification,
syould it defer automount until an actual attempt to open the file?
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
