I was pretty wary of AUTOIPL on our beloved production sysplex for quite a 
while but finally caved to best-practice arguments. IBM's long-held position is 
that a dead system is bad for the whole sysplex. A comatose system may even be 
worse, as it still registers a heartbeat but has no brain function. The goal is 
to shoot a useless system and get it back up ASAP. It may seem reckless in view 
the agonizing hours we once spent trying to save a system that we finally had 
to IPL anyway. With the configuration I proposed in my earlier post, a dead 
system is partitioned out, SADed, and reIPLed all without waiting for the 
operator to return from the bathroom. 

I've chronicled here a case where we repeatedly lost a development system 
around 05:00. SAD and reIPL were so fast that operators did not notice. At that 
hour of the day, there were so few users (if any) logged on that no one called 
to complain. I grilled everyone to find out who had IPLed. No one fessed up. 
Eventually figured out it was AUTOIPL after the system ran out of real storage. 
(Early bug in R13.) 

So AUTOIPL requires an attitude adjustment for those of us in the biz for a 
long time. I believe it's worth the trepidation (Merriam Webster's word of the 
day). Plus it never hurts to bank another excuse for more red wine. ;-) 

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


-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf 
Of Elardus Engelbrecht
Sent: Tuesday, August 30, 2016 11:18 PM
To: [email protected]
Subject: (External):Re: SAD testing in sysplex

Jesse 1 Robinson wrote:

>(Maybe more than you wanted to hear about SAD)

Thanks for your SAD notes about this SAD state (pun intended, sorry, but I'm 
not feeling SAD. ;-D ), especially the useful notes about DIAGxx and AMDSADMP. 
This will help all of us in times of emergencies.


>... Depending on system use, this may all occur without operator 
>awareness(!)

This is the reason why we don't have AUTOIPL defined. Even the Health Checker 
check for AUTOIPL is turned off.

Also thanks for this note 'SYSPLEX' means 'using XCF'. It is indeed very true!

Thanks again.

Groete / Greetings
Elardus Engelbrecht


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

Reply via email to