> Q:
> is it non-disruptive or disruptive when massaging the Weight & number of
> Processors if do as follows?
In general it is non-disruptive. However big change, let's say from 75% to 5%
can be very troublesome.
- the Weight value: from HMC(2.14.1)>click on a CEC level>Operational
Customization>Change LPAR Controls, then update the Weight,
then click on "Change Running System" button, and let it take immediate
effective?
or, click on "Save and Change" button, does this take immediate effective as
well without any disruptive?
Note: this is to be done for a regular zOS lpar
Save and change make it effective and it will be saved for the future. What
future? POR.
- the number of Processors: Note: this is to be done for a CF lpar
from the HMC Operating System Message in CF lpar, issue D CPS and make sure
of the Processors status,
issue CP xx OFFLINE, then D CPS again to make sure of it taken offline,
from same path explained in first paragraph above, update the number of
Processors (this is Shared, not Dedicated in this case)
then click on "Change Running System" button, and let it take immediate
effective?
Note: this is a Shared Processor, thus other Lpars in this CEC will take
the advantage of Processors utilization as much as one processor
released from the CF lpar? and this must be non-disruptive right? No need
to move the Structures, and re-allocate them?
I admit, I don't understand the scenario. Maybe it is result of my poor
English, or maybe you can decribe it in simpler words.
Few remarks:
CF love dedicated engines! So any shared engine is good for non-prod
environment and even then people say it is not good idea. Note: things changed
recently, but I don't want to describe details, because my memory is failing in
this area.
In many cases single ICF is enough.
When changing CPU power, think about whole dedicated ICFs, not weights.
AFAIR for share ICF you can change the weights as usual. However last time I
did it 7 or 8 years ago.
Console commands (CF interface) regarding CPUs are non-disruptive (I believe
so), however change (remove) from SE is disruptive, unless things were changed.
In properly configured syplex you should have at least two CFs (no single point
of failure), so your mistake should trigger structure rebuild. It is not
something people like, but it is like failover procedure.
In case of doubt just do it first on sandbox sysplex.
- any others that I might miss at this moment?
Yes, but it depends on your needs and your setup.
For example, when you watch CPU weight change you will notice separate tabs for
GP (regular CP), zIIP, IFL, former zAAP. The tabs exist only if you have that
kind of processor.
It is general rule for HMC - you won't see things that are not enabled.
Sometimes enablement means configuration, and sometimes that means orderable
(usually paid) feature. And it depends on user authorities - SERVICE see other
things than SYSPROG. And PEMODE see a lot more.
HTH
--
Radosław Skorupka
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
