--- In email@example.com
, "somitcw" <somitcw@...> wrote:
> --- In firstname.lastname@example.org,
> "kerravon86" <kerravon86@> wrote:
> > I automatically generated the IEBCOPYs from the
> > existing batch file, BTW.
> Be careful with IEBCOPY until it's checked.
> If a PDS exceeds 32K tracks, IEBCOPY could have
> a problem.
Ok, but I am not allocating datasets that big.
Also the procedure I have failsafes, as the end
result can be verified against the PC.
> Test would be allocating a PDS as (CYL,(2184,,1))
Even on an empty 3390-2 I can't get 2184 cylinders.
It's probably a genuine error. I can actually get
more tracks than that - 32765 in fact. And that's
regardless of whether I put 1 directory block or
1000 directory blocks. Which is what it is meant
to be according to the manual - the space comes
from primary quantity, at least for allocation by
cylinder and track.
So a 3390-2 appears to be outside anyone's ability
to stuff up at a dataset level.
But let me try anyway.
What's interesting is that if I specify 32766, I
get a strange message:
IEF140I HERC01A S2 DD1 - DIRECTORY SPACE REQUEST LARGER THAN AMOUNT AVAILABLE ON THIS VOLUME
Actually it's the primary space that is too big.
There's plenty of space for the directory.
Let's switch to blocks, to make the directory
blocks really count.
32765 tracks is 98295 blocks of 18400.
After some experimentation, I came up with this:
//DD1 DD DSN=MVSSRC.EBCDIC,DISP=(,CATLG),
as the biggest number of directory blocks I could
specify. To get 75 directory blocks to work, I
needed to decrease the 98290 to 98289.
I decided to take a whole track off instead, and
went to 98287.
That then allowed me to push the directory blocks
up to 119. 120 crashed. So 1 track allowed me to
go from 74 to 119, an increase of 45. 45 directory
blocks per track is the expected number.
That initial 74 was probably 74-45=29 and a partial
allocation/initial stuffing around.
So the directory blocks in that 119 situation take
up 2 tracks, plus a bit of a 3rd. 98287 is 32762
tracks plus a bit of another. So in total that is
32762 + 2 + 1 = 32765 tracks.
The same figure that we had before, seemingly
You can't stuff up with a 3390-2 so easily. Not
even by trying to sneak in with directory blocks.
I'll try my real situation again.