I always do, and I always tell my customers NEVER do *IMMED. ALWAYS, use *CNTRLD. I have heard many CEs over the life of the AS/400 to IBM I, say the exact same thing as your post. I always do an ENDSBS *ALL *CNTRLD 60 with *NOJOBLOG. Then when the system is in a restricted state, I do a PWRDWNSYS *CNTRLD 30. Never had a problem doing this, but I have heard stories of people having problems with flushing the cache from the buffers onto the disks with *IMMED. Now, perhaps it isn't during the power down that the problem occurs, but what they are doing after they power the machine down.
You are right, *CNTRLD 30 becomes *IMMED in 30 seconds, but it gives the system 30 seconds to start to wrap up things and let the system come down. In addition, there are many jobs that receive the termination notice of either a PWRDWNSYS or ENDSBS and start shutting themselves down properly.
Perhaps this is belts and suspenders, but it sure beats having a problem.
From: midrange-l-bounces@xxxxxxxxxxxx [mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Paul Fenstermacher
Sent: Wednesday, January 16, 2013 4:37 AM
To: Midrange Systems Technical Discussion (midrange-l@xxxxxxxxxxxx)
A former co-worker had a problem last week during a scheduled maintenance window and the analysis afterward led a consultant to say this in his root cause analysis report:
"Anytime you are doing any kind of disk work on a system you always want to power the system down controlled. Ending the system "controlled" flushes the disk cache from memory to disk, "immediate" does not do this. I cringe when I see customers using *IMMED. 99% of the time the system will come back up without issue, but that other 1% results in the type of issue reported here. They took the time to end all of the subsystems to a restricted state before issuing the PWRDWNSYS, which is the right thing to do, but then powered down immediately. It's a common misconception that ending the system controlled takes significantly longer to end. Once the system is in a restricted state, it takes only seconds longer to end the system "controlled" versus "immediate". The trick to using *CNTRLD is to make sure you change the delay time to something like 5 (for 5 seconds) rather than the default of 3600, otherwise you'll be waiting an hour before the system starts powering down."
I have to say in 27 years of administering the platform in all its flavors I've never done a PWRDWNSYS *CNTRLD and I have yet to have any problems caused by that. Thoughts anyone??
Paul Fenstermacher | Sys/NW Admin,Sr | Corporate Systems - POWER Systems Administration | Jack Henry & Associates, Inc.(r)
663 West Highway 60 | Monett, MO 65708 | Ph. 417.235.6652 | x177389 | pfenstermacher@xxxxxxxxxxxxx<mailto:pfenstermacher@xxxxxxxxxxxxx>
NOTICE: This electronic mail message and any files transmitted with it are intended exclusively for the individual or entity to which it is addressed. The message, together with any attachment, may contain confidential and/or privileged information.
Any unauthorized review, use, printing, saving, copying, disclosure or distribution is strictly prohibited. If you have received this message in error, please immediately advise the sender by reply email and delete all copies.
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list To post a message email: MIDRANGE-L@xxxxxxxxxxxx To subscribe, unsubscribe, or change list options,
or email: MIDRANGE-L-request@xxxxxxxxxxxx Before posting, please take a moment to review the archives at http://archive.midrange.com/midrange-l