PDSE DELETE PENDING

Discussion:

PDSE DELETE PENDING

(too old to reply)

Kenneth J. Kripke

2019-08-29 05:05:44 UTC

The url
https://www.ibm.com/support/pages/pdse-member-pending-delete-processing-over
view indicates that when executing IEBPDSE with the
PARM='PERFORMPENDINGDELETE,NOANALYSIS'

Seems to imply that space will be released for deleted members. We do
routine maintenance on various Linklist datasets and follow-up with an LLA
UPDATE of the respective libraries on all LPARS. We have not tested the
IEBPDSE PERFORMPENDELETE,NOANALYSIS Yet, but will do so in the near future.
My question: How does this really work, and, what really has to happen to
reclaim the deleted space.

Sincerely;

Kenneth J. Kripke

***@comcast.net

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to ***@listserv.ua.edu with the message: INFO IBM-MAIN

Tabari Alexander

2019-08-29 11:39:28 UTC

Permalink

One of the differences between PDSes and PDSEs is that PDSEs reclaim the space of deleted members. In order for that to be done, however, the PDSE address space needs to know that the member is no longer in use so it is safe for the space to be reclaimed. If the space cannot be reclaimed at the time of deletion, which is typical if the PDSE is LLA managed, there are other opportunities where PDSE will try to reclaim the space as part of normal processing. IEBPDSE with the PerformPendingDelete parameter is a means to force the processing to occur. Please do note that this process will still only reclaim space from deleted members that are not in use.

Tabari

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to ***@listserv.ua.edu with the message: INFO IBM-MAIN

Steve Horein

2019-08-29 11:40:10 UTC

Permalink

I was unaware of the PERFORMPENDINGDELETE option. Nice.
Anecdotal: I'm a automation administrator, and have struggled with
NetView's DSILIST data set that is used to retain various automation table
listings/reports. IBM recommends allocating this data set as a PDSE to
avoid Sx37 ABENDs and/or the need to compress. Well, guess what?

IEC031I D37-04,IGG0210B,NETVIEW,NETVIEW,SYS89298,602D,SH9DAK,
NETVIEW.NET97.DSILIST(INGMRT01)

IEC217I B14-0C,IGG0201Z,NETVIEW,NETVIEW,SYS89298,602D,SH9DAK,
NETVIEW.NET97.DSILIST(INGMRT01)

I don't have the result of IEBPDSE, but if I recall, there were 83 pending
deletes at the time. This is on a test system, so nothing I was terribly
concerned with.
After reading your question, I exercised "PERFORMPENDINGDELETE" with the
following result:

IGW705I Pending Delete Records Processed
00000000 Pending Delete Members deleted
Out of 00000014 possible

What?! <expletive deleted>
NetView has a "REACC" command that does some magic in the background (
https://www.ibm.com/support/knowledgecenter/SSZJDU_6.2.1/com.ibm.itnetviewforzos.doc_6.2.1/dqc_reacc.htm).
Other doc references for the REACC command include: "If the data set
becomes full and you must compress the data set to add more command lists,
use the REACC command...".
I issued REACC and attempted "PERFORMPENDINGDELETE" again, with the
following result:

IGW705I Pending Delete Records Processed
00000014 Pending Delete Members deleted
Out of 00000014 possible

Much better.
While I can't help with the question of "what really happens...", it does
seem that -something- must happen to the PDSE to allow successful deletion
of those members. In the case of this particular NetView data set, "REACC"
appears to be that thing. Perhaps in the case of LLA managed data sets, it
would be the "F LLA,UPDATE=..."?

Post by Kenneth J. Kripke
The url
https://www.ibm.com/support/pages/pdse-member-pending-delete-processing-over
view indicates that when executing IEBPDSE with the
PARM='PERFORMPENDINGDELETE,NOANALYSIS'
Seems to imply that space will be released for deleted members. We do
routine maintenance on various Linklist datasets and follow-up with an LLA
UPDATE of the respective libraries on all LPARS. We have not tested the
IEBPDSE PERFORMPENDELETE,NOANALYSIS Yet, but will do so in the near future.
My question: How does this really work, and, what really has to happen to
reclaim the deleted space.
Sincerely;
Kenneth J. Kripke
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to ***@listserv.ua.edu with the message: INFO IBM-MAIN

Paul Gilmartin

2019-08-29 16:04:18 UTC

Permalink

Post by Steve Horein
Much better.
While I can't help with the question of "what really happens...", it does
seem that -something- must happen to the PDSE to allow successful deletion
of those members. In the case of this particular NetView data set, "REACC"
appears to be that thing. Perhaps in the case of LLA managed data sets, it
would be the "F LLA,UPDATE=..."?

After a program has issued BLDL or DESERV, PDSE is obliged to assume that
any members returned are in use until the program issues STOW DISC.
I surmise that REACC (sounds like CMS jargon) issues the DISC.

Does NetView gain any performance benefit by hoarding TTRs rather than
issuing DISC sooner?

-- gil

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to ***@listserv.ua.edu with the message: INFO IBM-MAIN

k***@gmail.com

2019-08-30 01:59:18 UTC

Permalink

Thank you to all that have responded to my questions. You have been most helpful regarding this matter.

Sincerely;

Kenneth J. Kripke
***@comcast.net