Hey Mr. Z,
I don't think we disagree. I'm just pointing out the increased risk of
making big changes implemented with a sysplex-wide IPL if there is another
way to achieve the goal. In particular, a sysplex cold start is about the
riskiest action you can take. Avoid it if you can.
My war story is so mundane it's almost laughable except that it crippled
the most important sysplex in the enterprise. Our beloved-cow-plex. 12
years ago I implemented the RACF STARTED class to replace the old load
module. Just a few months ago someone (else) made a truly throw-away change
to a RACF profile. Because of my screw-up, this change happened to hit the
userid that JES2 had been using for all 12 years. Following a total sysplex
outage just days after the profile change, JES2 abended S913 because he
could not update his own checkpoint on any system in the plex. The outage
was technically 'scheduled'--except for the actual up time hours later than
planned. During our usual rolling IPL, the same profile change would have
been mysterious and troublesome but easily fixed from another system.
Unfortunately RACF profiles are inaccessible from any system not connected
to the data base. A major owie.
.
.
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]
Mark Zelden
<[EMAIL PROTECTED]
CHNA.COM> To
Sent by: IBM [email protected]
Mainframe cc
Discussion List
<[EMAIL PROTECTED] Subject
.EDU> Re: Change CFRM policy name for IPL
09/05/2007 11:54
AM
Please respond to
IBM Mainframe
Discussion List
<[EMAIL PROTECTED]
.EDU>
On Wed, 5 Sep 2007 11:26:56 -0700, Skip Robinson <[EMAIL PROTECTED]>
wrote:
> A sysplex cold start is fraught with peril under the rosiest of
circumstances.
Obviously if you require 24 x 7 availability for applications, then you
don't
want (or schedule) a sysplex wide IPL. I'm also not advocating it to all
as a "best practice" even if you have the opportunity to do it. I was
offering
a another point of view (and a personal preference). So if availability
isn't
holding you back, the risks of regressing dynamic changes can be reduced by
IPLing more often.
The Pros/Cons of this are discussed in this redbook:
z/OS Planned Outage Avoidance Checklist
http://www.redbooks.ibm.com/abstracts/sg247328.html?Open
Do you run into problems with IPLs of single systems due to dynamic changes
being regressed? There is no inherent danger in doing a sysplex wide
IPL and I would say that to be as "afraid" of them as you appear to be,
then
perhaps the problem may be with change control procedures - not with doing
a
sysplex wide IPL.
BTW, the example I cited was the only one I could think of in my 5 years
here, but the timing was perfect since it just happened. It would have
been
better had it happened during a planned sysplex outage (which would have
been in few weeks anyway). I have seen the same issue with single
system IPLs though (I'm sure many of the subscribers have seen the
occasional dynamic APF library gone after an IPL).
I guess you have a war story of your own or you wouldn't have such
strong feelings. :-)
Mark
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
