Brent,
SRM still gets involved in volume selection for non-SMS volumes. It is a
much simpler algorithm, and my shaky memory says that the order is based on
LCU activity and number of allocated datasets on the volume.
Volumes already allocated to other new datasets in the same step also go to
the end of the queue - this stops you from having all your SORTWKS on one
volume.
SECONDARY space is not taken into consideration when the Primary space is
allocated.
Now what I don't understand about your problem description below is why you
have a problem when you code UNIT=(esoteric,6). Is this from an IEFBR14 at
the start of the job allocating a file to be written to in a subsequent
step? If so, your problem may be that non-SMS files do not catalog the UNIT
COUNT as unspecified candidate volumes.
For non-SMS you must also specify UNIT=(esoteric,6) in the JCL of the step
that writes the file so that SECONDARY extents can be allocated on up to 6
more volumes.
//ALLOC EXEC PGM=IEFBR14
//NEWFILE DD DSN=MY.NONSMS.DSN,DISP=(NEW,CATLG,DELETE),
// UNIT=ESOTERIC,
// SPACE=(CYL,(1000,500))
//STEP1 EXEC PGM=WRITEIT
//NEWFILE DD DSN=MY.NONSMS.DSN,DISP=OLD,
// UNIT=(ESOTERIC,6),
// ETC,ETC, ETC
In example above the UNIT COUNT can be omitted from the IEFBR14 as it is
never used. Having it in the step that writes to the file is where you need
it.
Ron
>
> For example, for a Physical Sequential file with
> UNIT=(esoteric,6)
> SPACE=(CYL,(1000,500))
> and 3390 geometry (3339 cylinders)
>
> we have observed DADSM pass over volumes with
> plenty of space and instead allocate 3 volumes
> that had preexisting files with 3000 cylinders.
>
> In other words, the most cylinders available
> would be 3339-3000 = 339 yet with a secondary
> of 500 cylinders DADSM would still pick the packs
> with less than 339 and the file would end
> up running out of space.
>
----------------------------------------------------------------------
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
