Kernelquestion

[Friedhelm Neyer]

Hi

this is a output from topas on a 9076-550 SP2-Node with 4 procs and 4 gb
memory
Cannot find out, why the kernel eats so much of the system


Wed Sep 24 12:57:49 2003 Interval: 2 Cswitch 1447 Readch
9893
Syscall 1236
Writech 3053
Kernel 74.6 |##################### | Reads 44 Rawin
0
User 5.1 |## | Writes 50 Ttyout
1771
Wait 2.3 |# | Forks 0 Igets
0
Idle 17.8 |##### | Execs 0 Namei
59
Runqueue 4.0 Dirblk
0
Interf KBPS I-Pack O-Pack KB-In KB-Out Waitqueue 2.1
en1 120.8 54.4 83.9 3.4 117.4
css0 4.0 5.9 4.9 2.9 1.1 PAGING MEMORY
Faults 241.4K
Real,MB 4095
Disk Busy% KBPS TPS KB-Read KB-Writ Steals 0 % Comp
69.9
hdisk1 20.9 78.9 18.4 0.0 78.9 PgspIn 0 %
Noncomp 30.8
hdisk0 16.9 78.9 18.4 0.0 78.9 PgspOut 0 %
Client 0.5
hdisk10 0.0 0.0 0.0 0.0 0.0 PageIn 0
hdisk6 0.0 0.0 0.0 0.0 0.0 PageOut 19 PAGING
SPACE
hdisk8 0.0 0.0 0.0 0.0 0.0 Sios 18 Size,MB
8192
% Used
13.8
vsdkp (26064) 3.6% PgSp: 0.0mb root % Free
86.1


The vmtune-output of the same machine is like this:

vmtune: current values:
-p -P -r -R -f -F -N -W
minperm maxperm minpgahead maxpgahead minfree maxfree pd_npages
maxrandwrt
52428 209713 2 96 480 576 524288 0

-M -w -k -c -b -B -u
-l -d
maxpin npswarn npskill numclust numfsbufs hd_pbuf_cnt lvm_bufcnt
lrubucket defps
838852 65536 16384 1 300 176 16 131072
1

-s -n -S -L -g -h
sync_release_ilock nokilluid v_pinshm lgpg_regions lgpg_size
strict_maxperm
1 0 0 0 0 0

-t
maxclient
209713

PTA balance threshold percentage = 50.0%

number of valid memory pages = 1048565 maxperm=20.0% of real memory
maximum pinable=80.0% of real memory minperm=5.0% of real memory
number of file memory pages = 318602 numperm=30.4% of real memory

number of compressed memory pages = 0 compressed=0.0% of real memory
number of client memory pages = 12 numclient=0.0% of real memory
# of remote pgs sched-pageout = 0 maxclient=20.0% of real memory


And last, the output from vmstat 1

kthr memory page faults cpu
----- ----------- ------------------------ ------------ -----------
r b avm fre re pi po fr sr cy in sy cs us sy id wa
7 2 854098 20186 0 0 1 615 613 0 381 339 223 26 23 47 4
3 2 854098 20184 0 0 0 0 0 0 882 1764 969 3 72 24 0
4 2 854098 20184 0 0 0 0 0 0 1180 1196 1502 5 72 23 0
3 3 854098 20184 0 0 0 0 0 0 934 890 1053 5 78 12 5
4 2 854098 20184 0 0 0 0 0 0 814 1040 869 6 72 22 0
3 2 854098 20183 0 0 0 0 0 0 754 718 752 3 72 24 0
3 2 854098 20183 0 0 0 0 0 0 771 727 727 4 74 18 4
3 2 854114 20167 0 0 0 0 0 0 828 2512 705 5 74 18 3
3 2 854098 20183 0 0 0 0 0 0 708 687 640 5 72 23 0
3 2 854103 20178 0 0 0 0 0 0 978 1139 1140 5 76 15 4
3 2 854098 20183 0 0 0 0 0 0 1443 2002 1871 6 73 22 0
3 2 854098 20183 0 0 0 0 0 0 1489 2735 1912 2 76 21 0
4 3 854098 20183 0 0 0 0 0 0 1273 1412 1336 6 75 15 4
3 2 854098 20183 0 0 0 0 0 0 986 884 1047 6 70 24 0
3 2 854098 20183 0 0 0 0 0 0 910 1030 953 4 72 24 0
3 2 854098 20183 0 0 0 0 0 0 1297 1350 1603 4 78 15 4
3 2 854098 20183 0 0 0 0 0 0 1669 1647 2494 3 76 22 0
3 2 854098 20183 0 0 0 0 0 0 1288 1077 1557 5 77 16 2
3 2 854098 20182 0 0 0 0 0 0 1027 1097 1257 5 71 24 0
3 2 854098 20182 0 0 0 0 0 0 902 850 963 4 72 24 0

Any hints how to decrease amount of kernel ?
Is there any param that can be changed ?
Which param should I change ?

Thx in advance

[Gary R. Hook]

Friedhelm Neyer wrote:

> Hi
>
> this is a output from topas on a 9076-550 SP2-Node with 4 procs and 4 gb
> memory
> Cannot find out, why the kernel eats so much of the system

Good grief, look at your paging space. You don't have enough
real memory for your workload. You're also experiencing 240,000
page faults per second. That's a bit, um, "aggressive".

Properly configure the box for the desired tasks and the system
overhead should take care of itself.

--
Gary R. Hook / AIX PartnerWorld for Developers / These opinions are MINE
__________________________________________________ ______________________