× The internal search function is temporarily non-functional. The current search engine is no longer viable and we are researching alternatives.
As a stop gap measure, we are using Google's custom search engine service.
If you know of an easy to use, open source, search engine ... please contact support@midrange.com.



I've been running with processor shared pool for about a week now.

3 core limit
2 lpars

Production, at peaks, hit it 2.0 limit
R&D, at peaks, hit 1.0 limit

What I'd like to do, if possible, is change/teak the two profiles so that if either LPAR can steel CPU from the other LPAR if CPU available.:

Production can steel CPU upwards of somewhere between 2 and 3.
R&D can steel CPU upwards somewhere between 1 and 3.

Is this possible?

Paul



-----Original Message-----
From: DrFranken [mailto:drfranken@xxxxxxxxxxxx]
Sent: Wednesday, March 14, 2018 1:49 AM
To: Steinmetz, Paul
Subject: Re: Increasing MAX CPU of LPAR - HMC managed



On 3/13/2018 8:19 PM, Steinmetz, Paul wrote:
Larry,

1) When I powered down all my LPARs to enable the profile changes for CPU sharing on next activation, did the server also need to be powered down to enable the shared processor pool?
I didn't bring the server down, only the LPARS.
Nope, yer good.
2) Also, I'm not seeing the same CPU averages on the WRKSYSACT screen compared to my MPG CPU graphs.
Still researching.
Not sure about that both should match here.
3) Both LPARs Average CPU rate showing 99.8 and 100.0.
These numbers seem high, not sure they are accurate.
This is the CPU Operating rate, NOT percent busy of the CPU. That is they are operating at rated speed and have not been de-clocked to save power/heat. No worries here.

Below you show CPUs in mid 70s utilization which is cool.
 - L

Production
Work with System Activity PENCOR05
03/13/18 11:59:44
Automatic refresh in seconds . . . . . . . . . . . . . . . . . . . 5
Job/Task CPU filter . . . . . . . . . . . . . . . . . . . . . . . . .10
Elapsed time . . . . . . : 00:00:02 Average CPU util . . . . : 73.7
Virtual Processors . . . . : 2 Maximum CPU util . . . . . : 75.1
Overall SQL CPU util . . . : .0 Minimum CPU util . . . . . : 72.3
Average CPU rate . . . . . : 99.9 Current processing capacity: 1.50

R&D
Work with System Activity PENCOR06
03/13/18 12:00:00
Automatic refresh in seconds . . . . . . . . . . . . . . . . . . . 5
Job/Task CPU filter . . . . . . . . . . . . . . . . . . . . . . . . .10
Elapsed time . . . . . . : 00:00:02 Overall CPU util . . . . : 60.5
Overall SQL CPU util . . . : .0
Average CPU rate . . . . . : 100.0
Current processing capacity: 1.00

-----Original Message-----
From: DrFranken [mailto:drfranken@xxxxxxxxxxxx]
Sent: Tuesday, March 13, 2018 11:02 AM
To: Steinmetz, Paul
Subject: Re: Increasing MAX CPU of LPAR - HMC managed

WRKSYSACT is your friend to start.

The 'Current Processing Capacity' shows the ASSIGNMENT to that partition. I'm looking at one with 0.10 CPU just now.

The %CPU total for that partition is based on that number.

So if a partition is at 100% CPU it is using 100% of that 0.1 If it's at 200% it's using 0.2 CPU.

If the partition has multiple cores available you'll see three numbers Average, Max and Min for the CPU for this partition.

Also on that screen is shown the 'Average CPU rate' which is the percentage of rated processor speed.  So on the system I'm looking at I see 60.9 which means it's pretty light just now.  If that number is routinely at 100 then you're pushing the system fairly hard. Note that this number is for the Hardware so it's ALL partitions.

You can also enable data collection for a partition at the HMC level and then performance tools can collect that data.

Some of the sessions that Dawn May and others do at COMMON Conferences show how to view and graph all of that.

    - L



On 3/13/2018 10:42 AM, Steinmetz, Paul wrote:
Larry,

Processor pool is now enabled.

How can one tell how much CPU is being used by each LPAR at any given moment?

Paul

-----Original Message-----
From: DrFranken [mailto:drfranken@xxxxxxxxxxxx]
Sent: Friday, March 02, 2018 3:45 PM
To: Steinmetz, Paul
Subject: Re: Increasing MAX CPU of LPAR - HMC managed

On 3/2/2018 3:41 PM, Steinmetz, Paul wrote:
1) Do you think we will ever see the day that you can change and activate the profile changes without an IPL?
Yep. Note sure when though. Pretty deep in firmware to change that and the O/S cares too!
2) Are you saying the rules change from P7+ and above?
Yes. Minimum processor per LPAR is 0.05 on P7+ "D" models and newer. So the minimum PER VP is also 0.05 processor.
3) I redid the config. I tried dropping the .0 on the VP, but they get replaced.
Told you, IBM's interface is broken. :-)  But try 2.5 it won't allow that will it. LOL!

    - Larry
Minimum shared processing units: 0.3
Desire shared processing units: 1.5
Maximum shared processing units: 3.0
Shared processor pool: SharedPool01(1)

Virtual processors.
Minimum processing unites required for each virtual processor: 0.10
Minimum virtual processors: 1.0
Desired virtual processors: 2.0
Maximum virtual processors: 3.0

Sharing mode
X Uncapped Weight: 128

Paul

-----Original Message-----
From: DrFranken [mailto:drfranken@xxxxxxxxxxxx]
Sent: Friday, March 02, 2018 3:21 PM
To: Steinmetz, Paul
Subject: Re: Increasing MAX CPU of LPAR - HMC managed

Rules about Virtual processors (VPs):

They are always whole units, hence the 'rounding up'.

The maximum VPs must be higher than the Maximum processor. So if 2.50 maximum processor then maximum VP must be 3. It can be higher still but normally would not be.

For each VP assigned at the current time at least 0.1 (for POWER7)
or
0.5 (for POWER7+ & newer) processors must be assigned. So if you
have a
POWER7 and 3 VPs currently assigned the MINIMUM processor you can assign is 0.30.

This is how they get the 10, it's the ratio for POWER7, they just make it hard to understand with their example!!

For the pool do not 'reserve' any processor just set the limit to 3 processors.

Note that for VPs I always write 2 not 2.0 as 2.0 indicates you
COULD do 2.5. It's because my wife and son are chemists and the
inclusion of an additional decimal position indicates accuracy TO
THAT position. We cannot do that so I write 2. IBM's displays are wrong.
:-)

Now with your config understand the 'Desired VPs of 2'. You have uncapped (good) and a maximum of 3. When the partition starts it will get 1.50 processors and 2 VPs. Being uncapped it can then beg/borrow/steal UP TO 2.00 processors. The limit being that there are only 2 VPs so it can't go beyond that. Now if the VPs were desired at 3 then you could beg/borrow/steal up to 3.00 processors.

I'm not saying what is right for you just pointing out the difference.

As you show it if you wanted to go beyond 2.00 processors then you would dynamically up the VPs to 3 with the HMC.

Make sense?

- Larry


On 3/2/2018 11:25 AM, Steinmetz, Paul wrote:
Larry,

On my production lpar, here's my new profile settings.
I will be changing from capped to uncapped, and now using Shared processor pool 1.

I'm not sure on the rules for the virtual processors.
Should virtual equal processing unit?
They always seem to round up to the next higher integer.
I'm not totally understanding the help text, see below.
In their example, I don't see how they get 10.
When I try putting in a higher number, states invalid.
Also, what is a good recommendation for weight?

Desired virtual processors

The desired number of virtual processors for the logical partition that is associated with this partition profile. The desired number of virtual processors is the number of virtual processors that you want to assign to this logical partition when you activate this partition profile.

When you activate this partition profile, the managed system
calculates the ratio between the number of processing units
committed to the logical partition and the desired number of virtual processors.
If the calculated ratio is equal to or greater than the minimum
processing units required for each virtual processor, then the
logical partition is assigned the desired number of virtual processors.
Otherwise, the number of virtual processors is the number of
processing units divided by the minimum processing units required
for each virtual processor, rounded down to the previous whole number.
For example, a logical partition has 1.00 processing units
committed to it, and the minimum processing units required for each
virtual processor is 0.1. If the desired number of virtual
processors is less than 10, then the managed system uses the
desired number of virtual processors for this logical partition.
Otherwise, the number of virtual processors for this logical
partition is 10 (1

.00 processing units divided by a minimum of 0.1 processing units for each virtual processor).

Any change you make on this page takes effect only after you shut down the logical partition and reactivate this partition profile.

Minimum shared processing units: 1.5
Desire shared processing units: 1.5
Maximum shared processing units: 3.0
Shared processor pool: SharedPool01(1)

Virtual processors.
Minimum processing unites required for each virtual processor: 0.01
Minimum virtual processors: 2.0 Desired virtual processors: 2.0
Maximum virtual processors: 3.0

Sharing mode
X Uncapped Weight: 128

Paul

-----Original Message-----
From: MIDRANGE-L [mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf
Of DrFranken
Sent: Thursday, March 01, 2018 3:09 PM
To: Midrange Systems Technical Discussion
Subject: Re: Increasing MAX CPU of LPAR - HMC managed

YES you CAN set all partitions for a high of 3. The IBM i LPARS should all be set to draw out of a shared processor pool with a 3 core limit.
That will keep you clean.

Nothing else should draw from this pool. Do not use the defaultpool.

- Larry "DrFranken" Bolhuis

www.Frankeni.com
www.iDevCloud.com - Personal Development IBM i timeshare service.
www.iInTheCloud.com - Commercial IBM i Cloud Hosting.

On 3/1/2018 3:01 PM, Steinmetz, Paul wrote:
I have 8 active cores.
3 cores licensed for i5/OS.
Can I set the high max to 3 for all LPARs.
I thought I remember an error occurring if the total max exceeded the total licensed?
This was back in 2012, so I don't have the details.

Paul

--
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,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx Before posting, please take a moment to review the archives at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related questions.

Help support midrange.com by shopping at amazon.com with our
affiliate
link: http://amzn.to/2dEadiD

--
Chief mad-scientist
Frankeni Technology Consulting, LLC
Middleville, MI 49333
(616) 855-1667 - Office
(616) 260-4746 - Mobile

www.frankeni.com

lbolhuis@xxxxxxxxxxxx


--
Chief mad-scientist
Frankeni Technology Consulting, LLC
Middleville, MI 49333
(616) 855-1667 - Office
(616) 260-4746 - Mobile

www.frankeni.com

lbolhuis@xxxxxxxxxxxx


As an Amazon Associate we earn from qualifying purchases.

This thread ...

Follow-Ups:
Replies:

Follow On AppleNews
Return to Archive home page | Return to MIDRANGE.COM home page

This mailing list archive is Copyright 1997-2024 by midrange.com and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available on our policy page. If you have questions about this, please contact [javascript protected email address].

Operating expenses for this site are earned using the Amazon Associate program and Google Adsense.