Who holds up page data sets
Victor Zhang
We have a system with less 500M central memory installed,and it runs db2,7-8 cics regions,and tcp/ip,tso and we defined 5 local page data sets ,with each 300 cyls allocated,when all address spaces start up,the local page data sets are used 30% or above,does anyone here know if there is a method to know which address spaces holds up the page data sets,we can't see much demand paging by printing rmf report.
Thanks
Victor
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to list...@bama.ua.edu with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
Mark Zelden
> We have a system with less 500M central memory installed,and it runs
>db2,7-8 cics regions,and tcp/ip,tso and we defined 5 local page data sets,
>with each 300 cyls allocated,when all address spaces start up,the local
>page data sets are used 30% or above,does anyone here know if there is a
>method to know which address spaces holds up the page data sets,we can't
>see much demand paging by printing rmf report.
>
I am at home right now and can't remember for sure, but the RMF III
STORF command may show aux slots. It shows central and expanded frames
in use.
Mark
+------------------------------------------------------------------+
| Mark Zelden - OS/390 Consultant: 3D Business Solutions, Chicago |
| mailto:mze...@flash.net mailto:mze...@3dsolutions.com |
| Mark's MVS Utilites: http://home.flash.net/~mzelden/mvsutil.html |
+------------------------------------------------------------------+
Steve Samson
>
> Hi all,
> We have a system with less 500M central memory installed,and it runs db2,7-8 cics regions,and tcp/ip,tso and we defined 5 local page data sets ,with each 300 cyls allocated,when all address spaces start up,the local page data sets are used 30% or above,does anyone here know if there is a method to know which address spaces holds up the page data sets,we can't see much demand paging by printing rmf report.
>
> Thanks
Victor,
Not sure what you mean by "holds up" the locals, but you are operating
them too full. After system startup, page data set occupancy should not
exceed 25% or else the contiguous slot allocation algorithm (block
paging) will start to fail. Above 30%, the paging I/O is far less
efficient than it would be below 25%. I'd recommend adding one or two
more data sets of the same size.
Steve
Norman Hollander
PLPA 1 Cylinder (yes that is what you are seeing- let it overflow to CSA
since there is virtually no
activity on it after IPL. Put it on the
same volume as CSA)
CSA 150-200 Cylinders
Locals the entire 3390-3 volume (I like having 7-10 volumes; you could
have less but have some
available for emergency PageAdds.
Who is using your Central and Expanded Storage cuasing the system to Page
(either a large number
of pages between Central and Expanded, or Migration to Aux)? Check DB2
Bufferpools. Creative
DBAs reason that if they need 5,000 buffers then specifying 50,000 is
better. Multiply this by the
number of Bufferpools (because you must have a separate pool for 1k, 2k, 3,
4k, 8, 16k, 32k, etc,
even if they are not in use- NOT!) The bufferpools are allocated and fixed,
so no one else can use
it. I prefer to look at the min and max of the needed buffers at a peak
time and start with maybe 2-3
times the amount actually used. CICS will be much happier if it doesn't
need to page. OM/DB2 will
show the Bufferpool usage, and it is realtime dynamically changeable.
Edward Gould
>PLPA 1 Cylinder (yes that is what you are seeing- let it overflow to CSA
>since there is virtually no
> activity on it after IPL. Put it on the
>same volume as CSA)
>CSA 150-200 Cylinders
>Locals the entire 3390-3 volume (I like having 7-10 volumes; you could
>have less but have some
> available for emergency PageAdds.
Norm,an,
In light of the opinions discussed earlier this year on the list.
Small PLPA no longer makes sense.
Ed
Thomas Conley
>
> > We have a system with less 500M central memory installed,and it runs
> >db2,7-8 cics regions,and tcp/ip,tso and we defined 5 local page data
sets,
> >with each 300 cyls allocated,when all address spaces start up,the local
> >page data sets are used 30% or above,does anyone here know if there is a
> >method to know which address spaces holds up the page data sets,we can't
> >see much demand paging by printing rmf report.
> >
>
Vic,
The "Real" field on the SDSF DA screen shows you how much storage is held by
each started task.
Regards,
Tom Conley
Victor Zhang
Victor
___________________________________________IP卡、上网卡两折史奴比、蜡笔小新全家福华语排行榜颁奖典礼加入书友会获精美礼品
Victor Zhang
Edward E. Jaffe
>
> The "Real" field on the SDSF DA screen shows you how much storage is held by
> each started task.
Tom, the "Real" column merely shows how many central storage frames are
currently assigned to the address space. It does not show how much
virtual storage the address space uses, nor how many auxiliary slots
have been assigned.
--
-----------------------------------------------------------------
| Edward E. Jaffe | |
| Mgr, Research & Development | edj...@phoenixsoftware.com |
| Phoenix Software International | Tel: (310) 338-0400 x318 |
| 5200 W Century Blvd, Suite 800 | Fax: (310) 338-0801 |
| Los Angeles, CA 90045 | http://www.phoenixsoftware.com |
-----------------------------------------------------------------
Sam Knutson
<ISV plug mode on>
TMON for MVS from Landmark
option 1.7.1 AUXILIARY STORAGE STATUS
shows this by #frames, %, and has the information for each
job and at a summary level.
<ISV plug mode off>
Take a look at
http://home.flash.net/~mzelden/mvsfiles/irampf.txt
which is an MPF exit written by Shane Ginnane and discussed on
IBM-MAIN.
Search the archives
http://groups.google.com/groups?oi=djq&as_ugroup=bit.listserv.ibm-main
using keyword IRAMPF to get most of the thread. I have not tried it but
at a glance it looks like you should be able to run this is a batch program
just needs to be authorized. You could probably easily modify the code
to elminate the WTOR processing and just print out the full report.
Jim Mulder of IBM made an excellent post that summarized what
was needed to accurately report on this. IIRC this was the basis for IRAMPF but
I could be dead wrong about that as I haven't had any coffe yet this morning
and the neurons are not jumping around real quick:-)
-----------------------
I had suggested some years ago that we should provide a
D ASM,USERS(nn)
to display the nn largest consumers of Aux Storage. Looks like it
didn't happen. Feel free to submit a requirement. The MPF exit
you suggest is fairly easy to write. Simply obtain the local and CMS
locks, and for each in-use slot in the ASVT, the ASCB points to an ASSB.
ASSBVSC is the number of VIO Aux pages the address is using, and
ASSBNVSC is the number non-VIO Aux pages. ASSBJBNS or ASSBJBNI
has the job name. Loop though the ASVT saving this data for the nn
address spaces with the highest (ASSBVSC + ASSBNVSC), release the
CMS and local locks, and WTO the results.
Jim Mulder z/OS System Test IBM Corp. Poughkeepsie, NY
----------------------
I hope that helps.
Thanks,
Sam Knutson
Landmark Systems Corporation http://www.landmark.com
mailto:sknu...@landmark.com Mail me at work
My Personal Web Site: http://www.knutson.org Odds & Ends
CBT Tape Home Page: http://www.cbttape.org MVS Freeware
Ginnane, Shane
Have replied privately to Victor suggesting he pull IRAMPF.
Yes, it runs fine as a batch run - just remember to warn your operators
first !!.
Percentage code will be pretty simple to implement - when I get auth to
submit stuff to CBT I'll include this.
Checking for batch runs likewise - again I'll look at cutting some code for
that.
I think I got all of Peters' suggestions in it.
Shane ....
-----Original Message-----
From: Sam Knutson [mailto:sknu...@LANDMARK.COM]
Take a look at
http://home.flash.net/~mzelden/mvsfiles/irampf.txt
which is an MPF exit written by Shane Ginnane and discussed on
IBM-MAIN.
Search the archives
http://groups.google.com/groups?oi=djq&as_ugroup=bit.listserv.ibm-main
using keyword IRAMPF to get most of the thread. I have not tried it but
at a glance it looks like you should be able to run this is a batch program
just needs to be authorized. You could probably easily modify the code
to elminate the WTOR processing and just print out the full report.
**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager of QR.
This message has been swept by MIMESweeper for the presence of computer
viruses. No warranty is given that this message upon its receipt is
virus free and no liability is accepted by the sender in this respect.
This email is a message only; does not constitute advice and should not
be relied upon as such.
**********************************************************************
EXT / B&M BARNARD-BROWN David
list out in decending order all users of Auxilary Srorage.
/* REXX */
/*--------------------------------------------------------------------*/
trace 0
inc = 1
address tso
numeric digits 10
cvt = c2d(storage(10,4))
cvtasvt = c2d(storage(d2x(cvt+556),4))
asvtmaxu = c2d(storage(d2x(cvtasvt+516),4))
asvtenty = c2d(storage(d2x(cvtasvt+528),4))
do cntr = 0 to asvtmaxu
offset = cntr * 4
in_use_flag = storage(d2x(cvtasvt+528+offset),1)
if bitand(in_use_flag,'80'x) = '80'x then iterate
ascb = c2d(storage(d2x(cvtasvt+528+offset),4))
if ascb = '0' then leave
call get_assb
end
maxlines = inc - 1
outline.0 = maxlines
call Sort
say " Auxilary Storage Usage "sysn
say "Init. Job/STC/UserId. VIO Non-VIO Total"
do inc = 1 to maxlines
say outline.inc
end
exit
/*------------- get VIO and non-VIO page usage -----------------------*/
get_assb:
assb = c2d(storage(d2x(ascb+336),4)) /* address of ASSB */
assbvsc = c2d(storage(d2x(assb+32),4)) /* VIO Aux pages used */
assbnvsc = c2d(storage(d2x(assb+36),4)) /* non-VIO Aux pages */
assbjbni = storage(d2x(assb+336),8) /* initial jon name */
assbjbns = storage(d2x(assb+344),8) /* job name */
vio_pages = right(assbvsc,8)
non_vio_pages = right(assbnvsc,8)
total_aux = right(assbvsc + assbnvsc,8)
init_name = right(assbjbni,8)
job_name = right(assbjbns,8)
outline.inc = init_name' 'job_name' 'vio_pages' 'non_vio_pages' 'total_aux
inc = inc + 1
return
/*------------------ sort descending ---------------------------------*/
Sort: Procedure Expose outline.
do pntr1 = outline.0 to 1 by -1 until sort_done = 1
sort_done = 1
do pntr2 = 2 to pntr1
pntr3 = pntr2 - 1
if substr(outline.pntr3,36,8) << substr(outline.pntr2,36,8) then
do
xchg_field = outline.pntr3
outline.pntr3 = outline.pntr2
outline.pntr2 = xchg_field
sort_done = 0
end
end
end
return
-----Message d'origine-----
De: Sam Knutson [mailto:sknu...@LANDMARK.COM]
Date: lundi 10 septembre 2001 14:03
À: IBM-...@BAMA.UA.EDU
Objet: Re: Who holds up page data sets
Hi,
<ISV plug mode on>
TMON for MVS from Landmark
option 1.7.1 AUXILIARY STORAGE STATUS
shows this by #frames, %, and has the information for each
job and at a summary level.
<ISV plug mode off>
Take a look at
http://home.flash.net/~mzelden/mvsfiles/irampf.txt
which is an MPF exit written by Shane Ginnane and discussed on
IBM-MAIN.
Search the archives
http://groups.google.com/groups?oi=djq&as_ugroup=bit.listserv.ibm-main
using keyword IRAMPF to get most of the thread. I have not tried it but
at a glance it looks like you should be able to run this is a batch program
just needs to be authorized. You could probably easily modify the code
to elminate the WTOR processing and just print out the full report.
Jim Mulder of IBM made an excellent post that summarized what
I hope that helps.
Thanks,
Sam Knutson
----------------------------------------------------------------------
Eric Bielefeld
Eric Bielefeld
Sr. MVS Systems Programmer
P&H Mining Equipment
Milwaukee, WI
414-671-7849
eb...@phmining.com
>>> ext.b_m.ba...@SNCF.FR 09/13/01 01:08AM >>>
----------------------------------------------------------------------