Hi Peter,
Since SCSQAUTH contains all MQ LNKLSTd modules (and no MQ modules are in
any other LNKLSTd PDS(e)) and MQ, CICS, IMS and Batch are down, I can
guarantee that no User/Task/Job will attempt to fetch an MQ module.
There are 2 reasons why I would not compress SCSQAUTH:
1) It's a PDSE (not a PDS). My understanding (please correct me if I'm
wrong) is that PDSEs do not need compression to reclaim space left
behind by deleted/replaced members.
2) The Dataset has every member deleted. Afterwards, I IEBCOPY into it.
Why would anyone compress an empty dataset (or a recently populated one)?
Stopping LLA ... that affects the customer's (not robust as it should
be) Automation. Our implicit rule is that the fewer tasks involved, the
greater chance of success (implementation and staying within the
maintenance window).
The "real world" often has considerations that you won't find in a
pristine laboratory setting. (I've worked many years in both.)
Regards,
David
On 2020-10-20 09:14, Peter Relson wrote:
<snip>
How about this situation ...
I am part of a team of people who plan maintenance upgrades many months
in advance.
There is no possibility of IPL (for many of the maintenance upgrades).
All Batch is held (other than implementation jobs); DFSMShsm, CICS, IMS
and DB2 are down.
TSO is limited to implementers.
SCSQAUTH (an MQ LNKLSTd Dataset) is emptied, its contents replaced and
LLA is REFRESHd.
MQ, CICS, IMS and DB2 are started and tested by implementers and then by
the customer.
After everyone signs off, it's back to "business as usual" and Batch is
restarted.
Do you see any holes in this approach?
</snip>
If you can guarantee that no one does any fetch that might involve a
member with the same name as one within the PDS that you emptied then
"maybe".
And if you don't need to compress the data set then "maybe".
The operating system cannot stop you from compressing the data set. But it
tries a little, by allocating the data set with DISP=SHR, so that if you
do what you really should do for a compress, of using DISP=OLD, you won't
be able to.
If you're going to refresh LLA anyway, why not be safer and stop LLA
first, then restart when you're ready? The refresh will negate any
relevant caching of directories or modules, so there's no real benefit to
having LLA up across the operations.
Peter Relson
z/OS Core Technology Design
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
