I want to reiterate an endorsement of AUTOIPL. It costs nothing but some 
configuration time. In DIAGxx you put 

   AUTOIPL SADMP(dddd,SMSYSC) MVS(LAST)  /* IPL SAD FROM SAD IPL vvvvvv */

When we first went to z/OS R13, we stumbled on an obscure memory leak 
problem that took IBM a while to figure out. We came in more than once in 
the morning to find that our development system had IPLed at 
oh-dark-thirty and had a fresh SAD from just before IPL. The system had 
run completely out of real storage, wait stated, took SAD according to the 
statement in DIAGxx, and reIPLed itself. Even Ops didn't always notice.

There's one caveat to add. Not long ago we went through a 'DASD refresh' 
that moved every volume to a different address. At IPL time, z/OS checks 
the specified SADMP address for validity. If he determines that it's 
invalid--in our case the address was no longer genned--then the entire 
statement is discarded. Not just no SADMP but no AUTOIPL. Takes another 
IPL with correct values to fix it. Tough to remember, but in case it ever 
happens to you...

.
.
JO.Skip Robinson
Southern California Edison Company
Electric Dragon Team Paddler 
SHARE MVS Program Co-Manager
626-302-7535 Office
323-715-0595 Mobile
[email protected]



From:   "[email protected]" <[email protected]>
To:     [email protected], 
Date:   01/21/2013 10:48 PM
Subject:        Re: Stand-alone Dump Revisited
Sent by:        IBM Mainframe Discussion List <[email protected]>



> I saw Barbara's note about including the dataspaces belonging to OMVS, 
but haven't seen a "complete" generation deck (or significant excerpt) 
from her yet.

You've asked for it:
         AMDSADMP IPL=D3390,VOLSER=xxxxxx,                             +
               REUSEDS=ALWAYS,IPLEXIST=YES,                            +
               DUMP=('DSP OF ASID(ALL) ALSO PAGETABLES OF DATASPACES   +
               ALSO SP(ALL) IN ASID(ALL)                               +
               ALSO HIGH VIRTUAL IN ASID(ALL)'),                       +
               MINASID=ALL,                                            +
               DDSPROMPT=NO,                                           +
               OUTPUT=(Dyyyy,MVSR.SADMP),                              +
               CONSOLE=SYSC
         END

We had been running with this and even taking a few sadumps. A heavily 
(over-)used 8GB lpar took about 10 minutes to 3 volsers via 
autoipl/sadump. The beauty of the above is that it works on autoipl with 
no operator intervention required (you know, those 30 minutes operators 
take to find the correct document, read through it and do what it says?). 
The MVSR.SADMP is a historical relict and had caused me some grief on an 
sadump restart. Check the archives, at that time we discussed why it would 
be better to use sys1.sadmp.

And I see that we were better than system test - we did specify HIGH 
VIRTUAL :-) I just wanted to avoid software supports' constant 'The data 
are not in the dump' arguments by specifying everything and then proving 
to them that they have no clue where to look. So I made it easy for me and 
just specified ALL for just about everything. 

Barbara



----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to