On Tuesday, 05/24/2011 at 05:55 EDT, Mike Walter
<mike.wal...@aonhewitt.com> wrote:
> I have seen similar shutdown situations here - no doc regarding why the
> system went down, just guesswork.
>
> Our old z800 system running z/VM 5.4.0 was properly SHUTDOWN REIPLed
> Sunday evening and responds now with:
> cp q shutdowntime
> System shutdown time: 30 seconds; previous shutdown duration: 13 seconds
>
> The z10 running a different z/VM 5.4.0 system was DEACTIVATED from the
HMC
> (ignoring documented procedures) last Saturday as part of some hardware
> changes, is responds now with:
> cp q shutdowntime
> System shutdown time: 90 seconds
>
> Note that 90 seconds was allowed (since changed to 300 seconds), but
none
> was listed as "previous shutdown duration". That "previous shutdown
> duration" is documented as being reported "only if the system has been
> restarted after a system failure or restarted via a SHUTDOWN REIPL
> command."
>
> Couldn't an HMC DEACTIVATION be considered a "system failure", or is
> "system failure" meant as a CP ABEND causing such a failure so that CP
can
> determine the "previous shutdown duration"?
>
> Regardless, it would sure be nice if CP would issue a timestamped
message
> to the OPERATOR console when an HMC DEACTIVATE (or any other
> hardware-inspired shutdown) was occurring explaining what's happening
(and
> perhaps displaying the total current SHUTDOWNTIME value). If z/VM is
> responding to a SIGNAL SHUTDOWN from the hardware, it should still be up
> far enough to issue some sort of clarifying message. At least we'd know
> why there was no proper CP SHUTDOWN command, and can address that with
> (yet more) training. All we have now is a shrug of the shoulders and a
> 'gee, I don't know, someone else was on that shift'.
When the deactivation (aka shutdown or quiesce) signal is received, CP
will issue
HCPSHU6018I The processor controller has sent a shutdown signal
with a timeout interval of nnn seconds
to the OPERATOR's console (and probably emergency consoles). At this
point CP signals any enabled guests, waits for the configured amount of
time or for them all to shutdown, then finishes his own shutdown. Then
the LPAR is completely deactivated.
If the LPAR is deactivated with no delay, then there will be no message
since the blessed darkness comes instantly.
Whether or not you see a delay at LPAR deactivation depends on what
machine you have. The delay from the HMC used to be 0 (e.g. no delay) and
was changed in z10(?) to be hardcoded at 5 minutes. The z196 with HMC
2.11 may allow this value to be customized; I'm not sure.
Alan Altmark
z/VM and Linux on System z Consultant
IBM System Lab Services and Training
ibm.com/systems/services/labservices
office: 607.429.3323
mobile; 607.321.7556
alan_altm...@us.ibm.com
IBM Endicott
