Hi All,

We have a p570 system with 24 GB RAM and 6 CPUs. It is a database
server and the load is not high at the time the below screenshots are
taken.
Could someone help in figuring out why the paging space used is so much
inspite of no much activity? TIA. -- Ashok

# vmstat 1 10

System configuration: lcpu=12 mem=24576MB

kthr memory page faults cpu
----- ----------- ------------------------ ------------ -----------
r b avm fre re pi po fr sr cy in sy cs us sy id wa
0 0 2281919 7862 0 0 0 0 0 0 45 727 481 0 1 98 0
0 0 2281920 7810 0 1 0 0 0 0 129 1915 890 2 1 97 0
0 0 2281920 7610 0 0 0 0 0 0 202 3209 1358 1 2 96 1
0 0 2281920 7608 0 0 0 0 0 0 80 528 572 0 1 99 0
0 0 2281920 7608 0 0 0 0 0 0 74 1415 493 2 1 97 0
0 0 2282321 7206 0 0 0 0 0 0 220 2079 719 1 1 97 1
0 0 2282321 7199 0 0 0 0 0 0 57 1169 474 0 1 99 0
1 0 2282353 7157 0 0 0 0 0 0 116 1076 544 1 1 98 0
0 0 2282565 6945 0 0 0 0 0 0 65 1749 505 1 1 98 0
0 0 2282563 6940 0 0 0 0 0 0 111 3246 910 1 1 98 0
# lsps -a
Page Space Physical Volume Volume Group Size %Used Active
Auto Type
paging00 hdisk25 pagingvg 4096MB 25 yes
yes lv
hd6 hdisk0 rootvg 2048MB 50 yes
yes lv

Ashok Sangra <aksangra@gmail.com> wrote:
> Could someone help in figuring out why the paging space used is so much
> inspite of no much activity? TIA. -- Ashok

Because the sleeping programs occupy a lot of memory.

Yours,
Laurenz Albe

Actually just because the current output of vmstat doesn't show any
paging activity, doesn't mean that the system isn't paging. It just
means that it's not paging right now. Any data that is paged out to
paging space will remain there until it is needed again, which could be
a long time!

I believe that the real problem you are running into is that you
typically need to tune AIX when it is used for a database server,
expecially if the database is filesystem based, since by default, AIX
will use all available memory to cache your database file, even though
they are being cached by the database. Depending on you AIX level,
there are different recommened ways of dealing with this
"double-caching" issue. The "old" way is to tune the "minperm" and
"maxperm" setting to limit the amount of RAM AIX will use for file
caching. The "new" way is you use some mount flags on the database
filesytems to tell AIX not to perform file caching on those
filesystems.

In order to go any further, I would need some further details. What
version of AIX, what database, is the DB filesystem or raw logical
volume based and the output of vmstat -v would help as well.

Hope this helps!

Ashok Sangra wrote:
> Hi All,
>
> We have a p570 system with 24 GB RAM and 6 CPUs. It is a database
> server and the load is not high at the time the below screenshots are
> taken.
> Could someone help in figuring out why the paging space used is so much
> inspite of no much activity? TIA. -- Ashok
>
> # vmstat 1 10
>
> System configuration: lcpu=12 mem=24576MB
>
> kthr memory page faults cpu
> ----- ----------- ------------------------ ------------ -----------
> r b avm fre re pi po fr sr cy in sy cs us sy id wa
> 0 0 2281919 7862 0 0 0 0 0 0 45 727 481 0 1 98 0
> 0 0 2281920 7810 0 1 0 0 0 0 129 1915 890 2 1 97 0
> 0 0 2281920 7610 0 0 0 0 0 0 202 3209 1358 1 2 96 1
> 0 0 2281920 7608 0 0 0 0 0 0 80 528 572 0 1 99 0
> 0 0 2281920 7608 0 0 0 0 0 0 74 1415 493 2 1 97 0
> 0 0 2282321 7206 0 0 0 0 0 0 220 2079 719 1 1 97 1
> 0 0 2282321 7199 0 0 0 0 0 0 57 1169 474 0 1 99 0
> 1 0 2282353 7157 0 0 0 0 0 0 116 1076 544 1 1 98 0
> 0 0 2282565 6945 0 0 0 0 0 0 65 1749 505 1 1 98 0
> 0 0 2282563 6940 0 0 0 0 0 0 111 3246 910 1 1 98 0
> # lsps -a
> Page Space Physical Volume Volume Group Size %Used Active
> Auto Type
> paging00 hdisk25 pagingvg 4096MB 25 yes
> yes lv
> hd6 hdisk0 rootvg 2048MB 50 yes
> yes lv

Thanks. Pls find details below:
AIX version : 5300-04-03
DB : Oracle 8i
DB is filesystem based
# vmstat -v
6291456 memory pages
5982761 lruable pages
7284 free pages
5 memory pools
841834 pinned pages
80.0 maxpin percentage
20.0 minperm percentage
80.0 maxperm percentage
73.9 numperm percentage
4425496 file pages
0.0 compressed percentage
0 compressed pages
74.4 numclient percentage
80.0 maxclient percentage
4453668 client pages
0 remote pageouts scheduled
1028 pending disk I/Os blocked with no pbuf
1160 paging space I/Os blocked with no psbuf
2740 filesystem I/Os blocked with no fsbuf
86 client filesystem I/Os blocked with no fsbuf
0 external pager filesystem I/Os blocked with no
fsbuf
-------------

In order to go any further, I would need some further details. What
version of AIX, what database, is the DB filesystem or raw logical
volume based and the output of vmstat -v would help as well.

jtrosky@gmail.com wrote:
> Actually just because the current output of vmstat doesn't show any
> paging activity, doesn't mean that the system isn't paging. It just
> means that it's not paging right now. Any data that is paged out to
> paging space will remain there until it is needed again, which could be
> a long time!
>
> I believe that the real problem you are running into is that you
> typically need to tune AIX when it is used for a database server,
> expecially if the database is filesystem based, since by default, AIX
> will use all available memory to cache your database file, even though
> they are being cached by the database. Depending on you AIX level,
> there are different recommened ways of dealing with this
> "double-caching" issue. The "old" way is to tune the "minperm" and
> "maxperm" setting to limit the amount of RAM AIX will use for file
> caching. The "new" way is you use some mount flags on the database
> filesytems to tell AIX not to perform file caching on those
> filesystems.
>
> In order to go any further, I would need some further details. What
> version of AIX, what database, is the DB filesystem or raw logical
> volume based and the output of vmstat -v would help as well.
>
> Hope this helps!
>
> Ashok Sangra wrote:
> > Hi All,
> >
> > We have a p570 system with 24 GB RAM and 6 CPUs. It is a database
> > server and the load is not high at the time the below screenshots are
> > taken.
> > Could someone help in figuring out why the paging space used is so much
> > inspite of no much activity? TIA. -- Ashok
> >
> > # vmstat 1 10
> >
> > System configuration: lcpu=12 mem=24576MB
> >
> > kthr memory page faults cpu
> > ----- ----------- ------------------------ ------------ -----------
> > r b avm fre re pi po fr sr cy in sy cs us sy id wa
> > 0 0 2281919 7862 0 0 0 0 0 0 45 727 481 0 1 98 0
> > 0 0 2281920 7810 0 1 0 0 0 0 129 1915 890 2 1 97 0
> > 0 0 2281920 7610 0 0 0 0 0 0 202 3209 1358 1 2 96 1
> > 0 0 2281920 7608 0 0 0 0 0 0 80 528 572 0 1 99 0
> > 0 0 2281920 7608 0 0 0 0 0 0 74 1415 493 2 1 97 0
> > 0 0 2282321 7206 0 0 0 0 0 0 220 2079 719 1 1 97 1
> > 0 0 2282321 7199 0 0 0 0 0 0 57 1169 474 0 1 99 0
> > 1 0 2282353 7157 0 0 0 0 0 0 116 1076 544 1 1 98 0
> > 0 0 2282565 6945 0 0 0 0 0 0 65 1749 505 1 1 98 0
> > 0 0 2282563 6940 0 0 0 0 0 0 111 3246 910 1 1 98 0
> > # lsps -a
> > Page Space Physical Volume Volume Group Size %Used Active
> > Auto Type
> > paging00 hdisk25 pagingvg 4096MB 25 yes
> > yes lv
> > hd6 hdisk0 rootvg 2048MB 50 yes
> > yes lv

This will help .. new tuning options on AIX ..

You used to have to lower maxperm / maxclient so that numperm was
always greater than maxperm / maxclient and then you would only page
out file pages, but now you have an option to do that for you :

http://www-03.ibm.com/support/techdo...Index/TD103148

HTH
Mark Taylor

Ashok Sangra wrote:
> Hi All,
>
> We have a p570 system with 24 GB RAM and 6 CPUs. It is a database
> server and the load is not high at the time the below screenshots are
> taken.
> Could someone help in figuring out why the paging space used is so much
> inspite of no much activity? TIA. -- Ashok
>
> # vmstat 1 10
>
> System configuration: lcpu=12 mem=24576MB
>
> kthr memory page faults cpu
> ----- ----------- ------------------------ ------------ -----------
> r b avm fre re pi po fr sr cy in sy cs us sy id wa
> 0 0 2281919 7862 0 0 0 0 0 0 45 727 481 0 1 98 0
> 0 0 2281920 7810 0 1 0 0 0 0 129 1915 890 2 1 97 0
> 0 0 2281920 7610 0 0 0 0 0 0 202 3209 1358 1 2 96 1
> 0 0 2281920 7608 0 0 0 0 0 0 80 528 572 0 1 99 0
> 0 0 2281920 7608 0 0 0 0 0 0 74 1415 493 2 1 97 0
> 0 0 2282321 7206 0 0 0 0 0 0 220 2079 719 1 1 97 1
> 0 0 2282321 7199 0 0 0 0 0 0 57 1169 474 0 1 99 0
> 1 0 2282353 7157 0 0 0 0 0 0 116 1076 544 1 1 98 0
> 0 0 2282565 6945 0 0 0 0 0 0 65 1749 505 1 1 98 0
> 0 0 2282563 6940 0 0 0 0 0 0 111 3246 910 1 1 98 0
> # lsps -a
> Page Space Physical Volume Volume Group Size %Used Active
> Auto Type
> paging00 hdisk25 pagingvg 4096MB 25 yes
> yes lv
> hd6 hdisk0 rootvg 2048MB 50 yes
> yes lv
>
Looking at your output - there is no paging taking place - either in or
out (pi and po). Looks as if the system is idle pretty much most of the
time.

Are you more concerned with the fact that your paging space has a high
%Used?