Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations SkipVought on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

AIX 5.3 tuning question

Status
Not open for further replies.
Apr 23, 2008
4
0
0
US
Sorry for the long message:

######### Server Info ############

System Model: IBM,9117-MMA
Processor Type: PowerPC_POWER6
Number Of Processors: 4
Processor Clock Speed: 4704 MHz
CPU Type: 64-bit
Kernel Type: 64-bit
LPAR Info: 1 greyhound
Memory Size: 16384 MB
Good Memory Size: 16384 MB

# oslevel -s
5300-06-02-0727

####### Changes we have been made to the server ######

ioo -p -o j2_dynamicBufferPreallocation=256
ioo -p -o j2_maxPageReadAhead=256
ioo -p -o j2_maxRandomWrite=32
ioo -p -o j2_nBufferPerPagerDevice=1024
ioo -p -o j2_nPagesPerWriteBehindCluster=64
ioo -p -o maxpgahead=16
ioo -p -o maxrandwrt=32
ioo -p -o numclust=8
ioo -p -o numfsbufs=2048
ioo -p -o pv_min_pbuf=2048
vmo -p -o lru_file_repage=0
vmo -p -o maxclient%=90
vmo -p -o maxperm%=90
vmo -p -o minperm%=5

MinServers - LEAST(#CPUs=8, 10)=8
MaxServers - (10*(#Disks=45)/#CPUs=8)=60

I configured page space as follows:

# lsps -a
Page Space Physical Volume Volume Group Size %Used Active Auto Type
paging00 hdisk3 pagingvg 8192MB 11 yes yes lv
paging01 hdisk2 pagingvg 8192MB 11 yes yes lv
hd6 hdisk1 rootvg 8192MB 10 yes yes lv

I also deactivated/activated all three page spaces.

################ Result ########################

Output from vmstat -v before the changes: (Uptime to this point ~180 days)

# vmstat -v | grep block
Mon Apr 7 19:13:00 EDT 2008
11624496 pending disk I/Os blocked with no pbuf
85953253 paging space I/Os blocked with no psbuf
2484 filesystem I/Os blocked with no fsbuf
0 client filesystem I/Os blocked with no fsbuf
41362018 external pager filesystem I/Os blocked with no fsbuf

After changing the parameters and rebooted the server: (Uptime to this point 10 days)

# vmstat -v | grep block
Tue Apr 22 11:42:13 EDT 2008
1267 pending disk I/Os blocked with no pbuf
13513203 paging space I/Os blocked with no psbuf
2484 filesystem I/Os blocked with no fsbuf
0 client filesystem I/Os blocked with no fsbuf
159202 external pager filesystem I/Os blocked with no fsbuf

As you can see, the "paging space I/Os blocked with no psbuf" is still increasing a lot. I'd like to know how to fix this.

Thank you!!!

Wei
 
What does "vmstat 1 20" and a "svmon -G" show when there is traffic on the box?
What does "svmon -P | grep -p Pid" say in the colum of Pgsp?
What kind of application is there running?
What does
Code:
svmon -P| grep -p Pid| grep -vE "^---|^$|Pid"| sort -rn -k 4,4| head -20
show?
And "vmo -x| grep lru" ?
Maybe increase numpsbufs by factor 2 - seemed to help with the blocked, but I am curious what kind of application is running there. Let's see the output of above commands or else you might consider to get more RAM.

laters
zaxxon
 
Thank you for the reply.

Here are the answers to your questions:

# vmstat 1 20

System configuration: lcpu=8 mem=16384MB

kthr memory page faults cpu
----- ----------- ------------------------ ------------ -----------
r b avm fre re pi po fr sr cy in sy cs us sy id wa
0 10 5387143 12567 0 1229 1215 31242 858344 0 3125 8890 10374 9 15 34 43
2 8 5387146 12799 0 1237 1268 23267 451644 1 2787 16098 16639 16 12 25 47
6 15 5387147 12690 0 1438 1365 18645 421601 2 3744 80561 62476 62 10 3 24
10 15 5387147 12462 0 1394 1091 18418 359688 0 3554 78019 61617 61 10 3 25
13 12 5387146 12648 0 1008 1144 20574 280842 0 2764 31393 24686 23 11 28 38
8 6 5387149 12757 0 1014 787 32204 394290 0 2706 15765 11364 10 15 34 41
7 10 5387144 13089 0 1274 1182 21401 596433 4 2866 36286 21253 23 11 29 36
4 16 5387145 12847 0 1321 1072 19548 541767 3 3676 83908 59858 64 11 3 23
7 12 5387144 12560 0 717 529 18196 612121 1 2799 88865 49529 55 10 10 25
2 3 5387143 12508 0 573 612 24285 502036 2 1682 16098 11762 12 12 41 35
0 6 5387146 12485 0 676 526 26766 1654604 1 1711 14585 9483 10 15 43 32
4 16 5393549 11970 0 401 5422 25740 669341 2 2639 10288 9711 9 13 31 46
10 11 5393776 12784 0 524 1472 12419 297682 0 2753 52774 28883 37 7 1 55
10 15 5393945 12473 0 526 578 19592 697213 0 2982 85080 54168 64 10 2 24
0 9 5387157 12995 0 704 388 12731 310770 0 2144 43777 30291 34 9 22 36
1 5 5387158 12476 0 554 78 23545 635804 2 1245 19030 9386 10 12 44 34
0 7 5387155 12672 0 589 0 25048 402879 0 1453 15908 9915 11 12 42 35
9 5 5387155 12653 0 685 0 20664 276262 1 2441 66095 36515 48 10 18 24
2 19 5387156 12470 0 535 0 18507 278081 3 2573 83352 50726 62 9 4 25
14 13 5387158 12579 0 203 0 23192 453023 1 1768 42228 28037 30 12 28 30

# svmon -G
size inuse free pin virtual
memory 4194304 4181479 12825 1961553 5389857
pg space 6291456 1783159

work pers clnt
pin 1961553 0 0
in use 3962918 0 218561

PageSize PoolSize inuse pgsp pin virtual
s 4 KB - 4056151 1747335 1880497 5247873
m 64 KB - 7833 2239 5066 8874

# svmon -P| grep -p Pid| grep -vE "^---|^$|Pid"| sort -rn -k 4,4| head -20
2200004 oracle 1434088 1421364 2975 1429709 Y N N
2187676 oracle 1434316 1421364 3100 1430004 Y N N
2154964 oracle 1434137 1421364 3092 1429864 Y N N
2040262 oracle 1434570 1421364 2986 1430197 Y N N
2023840 oracle 1641944 1421364 3174 1430953 Y N N
2003124 oracle 1522430 1421364 16854 1452718 Y N N
1994928 oracle 1533560 1421364 8619 1440523 Y N N
1978826 oracle 1448115 1421364 3222 1430258 Y N N
1958344 oracle 1448519 1421364 2997 1430410 Y N N
1949872 oracle 1533987 1421364 9759 1442783 Y N N
1937664 oracle 1551532 1421364 4985 1470174 Y N N
1929564 oracle 1640899 1421364 3439 1430254 Y N N
1913158 oracle 1447825 1421364 3071 1429790 Y N N
1880346 oracle 1434036 1421364 2903 1429571 Y N N
1847444 oracle 1523671 1421364 15985 1453563 Y N N
1843230 oracle 1514588 1421364 10318 1440512 Y N N
1827026 oracle 1554152 1421364 6943 1474286 Y N N
1716300 oracle 1434079 1421364 2815 1429515 Y N N
1659038 oracle 1532065 1421364 6423 1470174 Y N N
1634458 oracle 1434505 1421364 3677 1430767 Y N N

# vmo -x | grep lru
lru_file_repage,0,1,0,0,1,boolean,D,
lru_poll_interval,10,10,10,0,60000,milliseconds,D,
lrubucket,131072,131072,131072,65536,4194304,4KB pages,D,

The application we are running is database warehouse, ETL.

We do consider adding another 16 GB memory, it will be couple of weeks from now.

Thanks again!
 
One more piece of info (vmstat -v | grep block)

you can see how much the counter increased during the time this was running (30 minutes).


Wed Apr 23 12:01:00 EDT 2008
1267 pending disk I/Os blocked with no pbuf
15180098 paging space I/Os blocked with no psbuf
2484 filesystem I/Os blocked with no fsbuf
0 client filesystem I/Os blocked with no fsbuf
168121 external pager filesystem I/Os blocked with no fsbuf



Wed Apr 23 12:33:34 EDT 2008
1267 pending disk I/Os blocked with no pbuf
15243929 paging space I/Os blocked with no psbuf
2484 filesystem I/Os blocked with no fsbuf
0 client filesystem I/Os blocked with no fsbuf
168121 external pager filesystem I/Os blocked with no fsbuf


Thanks!
 
I have no idea what helpful parameter can be tuned VMM wise which you isn't already. I guess there is no other way than getting more RAM.

laters
zaxxon
 
Just curious to know what happened so far - did you find another way to tune or did you get your RAM yet?

laters
zaxxon
 
Just got the additional 16 GB memory, will see how it improves the performance...
 
I am no Oracle DBA but could it be that the size of the SGA might have been a bit too large compared to the installed RAM? I bet there are some recommendations how to size it. So I guess it should be not more than about 2/3 to 3/4 of your built in RAM, if it's running somewhat alone on that machine.

Just found this http://www.dba-oracle.com/t_sga_sizing.htm
He talks about "sga_max_size" and "sga_target" - maybe check these too.

laters
zaxxon
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top