Re: subsystem monitor -pool question. -- MIDRANGE-L

Thanks Jim, and all...

I have cleaned up my SBS descriptions, and they now look as expected.

I am attempting to clean up what pool management I have done so far.
I expect to upgrade my 8GB to 24GB shortly, but I do want to get *BASE
cleaned up...

On 1/24/2012 3:20 PM, Jim Oberholtzer wrote:

It almost looks like someone issued a command:
CHGSBSD SBSD(COMPILES) POOLS((1 *NOSTG))

I strongly suggest you:

CHGSBSD SBSD(COMPILES) POOLS((1 *BASE *N ))

To get the subsystem monitor running in *BASE.

Just for clarity I would :

CHGSBSD SBSD(COMPILES) POOLS((1 *BASE *N) (2 *SHRPOOL3 *N))

Then change all the routing entries to point to subsystem pool 2.

Stop and start the subsystem.

Then:

CHGSBSD SBSD(COMPILES) POOLS((4 *NOSTG *N))

Now you'll have that beast running as you want it to.

Jim Oberholtzer
Chief Technical Architect
Agile Technology Architects

On 1/24/2012 3:05 PM, Gqcy wrote:

/posting inline/

On 1/24/2012 2:51 PM, Jim Oberholtzer wrote:

> In my opinion your subsystem monitor jobs, particularly in the IBM
> supplied subsystems, SHOULD run in *BASE. There are a bunch of good
> reasons for it, but for now let's just say it's best practice.
>
> Each subsystem should have two memory pools assigned. *BASE as
subsystem
> pool 1 and *shrpoolxx for subsystem pool 2 ( or more as needed but
it's
> actually fairly rare ).
>
> Routing entries should all point to subsystem pool 2.
>
> Now in your example you have QHTTPSVR running in the same pools as the
> jobs. I like the concept of pushing the web server jobs into their own
> memory but the subsystem really should run in *BASE. That subsystem is
> going to be dealing with potentially 1000s of threads so in order to
> make sure the subsystem monitor job ALWAYS gets an activity level when
> it needs one, it should run in *BASE. If the subsystem monitor job is
> competing with the web servers for memory and activity level, and it
> needs to spawn another thread, it won't be able to if the activity
> levels are all used up. That will slow down all your web servers.
Hence
> keep it in *BASE.

this has been a problem... Thanks so much!

>
> I'm guessing in your COMPILES subsystem there are two subsystem memory
> pools. *BASE is 1 and *SHRPOOL3 is subsystem pool 4 (which unless you
> have 2 other subsystem pools does not make sense, although it does not
> hurt anything)

No, my COMPILES subsystem looks like:
Sbsd..........TtlStg...1...2...3...4...5
COMPILES.........00................12

I only have one pool entry in my SBSD.
all 3 routing entries point to pool 4.

>
> Keep in mind not to mix up:
> System pools: Assigned as the system starts up. Numbers available from
> WRKSYSSTS on the far left
> Shared Pools: A finite number of them all created by IBM. Assigned to
> each subsystem with CHGSBSD command.
> Subsystem pools: Again assigned with the CHGSBSD command.
>
> These three are different although they are all used to manage memory
>
> Jim Oberholtzer
> Chief Technical Architect
> Agile Technology Architects
>
>
> On 1/24/2012 2:28 PM, Gqcy wrote:

>> Why / What is causing some of my subsystem monitor jobs to run in
the
>> *base pool versus some of them are running in the *SHRPOOL I have
made.
>>
>> Example wrkactjob screen:
>> SBS/JOB.......Type....Pool
>> COMPILES.......SBS.....2
>> ..SCAN_JOB.....BCH.....4
>> ..COMPILE......BCH.....4
>> QHTTPSVR.......SBS.....8
>> ..WEBSERVER....BCH.....8
>> ..WEBSERVER....BCI.....8
>>
>> compiles sbsd:
>> poolid= 4, *SHRPOOL3
>> all rtg entries point to poolid 4
>>
>> QHTTPSVR sbsd:
>> poolid= 1, *SHRPOOL8
>> lone rtg entry points to poolid 1
>> --

--