system load average

[Teck Sim]

After I have changed timeslice from 1 to 10, and still both the load and idle are high.
I am running Oracle 7.3.4.4 database for HP-UX 11, and it's small data warehouse.

System: smmmx Thu Aug 15 14:16:58 2002
Load averages: 6.34, 7.01, 6.70
99 processes: 97 sleeping, 2 running
Cpu states:
LOAD USER NICE SYS IDLE BLOCK SWAIT INTR SSYS
6.34 0.2% 0.0% 0.6% 99.2% 0.0% 0.0% 0.0% 0.0%

Memory: 80380K (26340K) real, 94780K (38052K) virtual, 290128K free Page# 1/8

TTY PID USERNAME PRI NI SIZE RES STATE TIME %WCPU %CPU COMMAND
? 24011 oracle 156 20 10272K 552K sleep 5:10 5.66 5.65 ora_dbwr_SM
? 24031 oracle 156 20 14592K 4880K sleep 24:35 3.26 3.25 ora_snp4_SM
? 24013 oracle 156 20 10764K 664K sleep 2:49 1.66 1.65 ora_arch_SM
? 1230 www 154 20 12828K 4900K sleep 135:55 1.60 1.60 ns-httpd
? 25904 oracle 154 22 10828K 1112K sleep 6:48 1.17 1.17 oracleSMD2
? 28 root 152 20 0K 0K run 4:44 1.15 1.15 vxfsd
pts/tb 27053 tsim 168 20 716K 332K run 0:00 0.45 0.38 top
? 2 root 128 20 0K 0K sleep 0:27 0.28 0.28 vhand
? 1452 root 154 20 1468K 468K sleep 23:31 0.25 0.25 _upsd
? 3 root 128 20 0K 0K sleep 17:03 0.21 0.21 statdaemon
? 24035 oracle 156 20 10256K 500K sleep 4:04 0.14 0.14 ora_db01_SM
? 24015 oracle 156 20 10240K 520K sleep 3:13 0.12 0.12 ora_lgwr_SM
? 24037 oracle 156 20 10256K 500K sleep 3:58 0.10 0.10 ora_db02_SM

Any suggestion for the above top interpretation?
I don't have a glancePlus installed.

sar -q 5 10

HP-UX smmmx B.11.00 A 9000/803 08/15/02

14:28:03 runq-sz %runocc swpq-sz %swpocc
14:28:08 5.2 100 0.0 0
14:28:13 4.4 100 0.0 0
14:28:18 4.4 100 0.0 0
14:28:23 4.6 100 0.0 0
14:28:28 4.6 100 0.0 0
14:28:33 4.0 100 0.0 0
14:28:38 5.0 100 0.0 0
14:28:43 4.2 100 0.0 0
14:28:48 4.6 100 0.0 0
14:28:53 4.4 100 0.0 0

Average 4.5 100 0.0 0

Thanks.

[Uday_S_Ankolekar]

Try this command and see who is eating up your resources..

This will give you output of top 10 cpu intensive process.

UNIX95= ps -e -o "vsz pcpu ruser pid stime time state args" | sort -rn |head -10

- Goodluck,

-USA..

[John Downs_1]

If you sure nothing is causing it to go so high start looking at the patchs. I had the same problem when I first upgraded to 11.00. After adding the patchs recommanded by cadence (our primary application) the problem went away.

-john

[Sridhar Bhaskarla]

Hi,

What do your sar -d 5 10 and %wio in sar 5 10 say?

It may be because they are waiting on Disk IO.

-Sri

[Teck Sim]

sar -d 5 10

HP-UX smmmx B.11.00 A 9000/803 08/15/02

15:04:00 device %busy avque r+w/s blks/s avwait avserv
15:04:05 c0t6d0 1.40 0.50 1 19 3.79 13.98
c0t1d0 0.20 0.50 0 3 2.19 11.43
c0t3d0 0.40 0.50 0 6 5.51 11.90
c0t4d0 0.60 0.50 0 3 9.64 20.83
15:04:10 c0t6d0 1.40 0.50 2 6 4.32 7.99
c0t3d0 0.40 0.50 0 6 5.34 10.83
15:04:15 c0t6d0 0.40 0.50 0 4 9.81 16.52
15:04:20 c0t6d0 0.80 0.50 1 8 6.66 6.36
c0t3d0 1.20 0.50 1 13 6.81 11.30
c0t4d0 0.40 0.50 0 6 1.47 15.21
15:04:25 c0t3d0 0.80 0.50 1 10 5.26 12.46
c0t4d0 0.40 0.50 0 3 3.72 18.89
15:04:30 c0t6d0 0.80 0.50 1 13 4.88 11.30
c0t3d0 0.80 0.50 1 10 5.36 12.44
c0t4d0 0.60 0.50 0 6 5.98 14.32
15:04:35 c0t3d0 1.00 0.50 1 10 7.20 12.93
c0t4d0 0.20 0.50 0 3 2.05 12.72
15:04:40
15:04:45
15:04:50 c0t3d0 0.40 0.50 0 6 5.34 6.83
c0t4d0 0.40 0.50 0 6 6.58 11.82

Average c0t6d0 0.48 0.50 1 5 5.12 10.14
Average c0t1d0 0.02 0.50 0 0 2.19 11.43
Average c0t3d0 0.50 0.50 0 6 5.95 11.46
Average c0t4d0 0.26 0.50 0 3 4.83 15.02

[Sridhar Bhaskarla]

Hi (again),

Are you seeing any performance degradation on the system?.

Your sar -d is looking fine. The response times are slightly abnormal but I wouldn't worry about them as the disk usage and avqueue are less.

I would check up the patch level on the server and upgrade it to a reasonably recent level.

-Sri

[Mike Stroyan]

The issue is probably just top reporting a higher load average than it should. You could apply the following two patches that correct problems with the output of top-
PHCO_26020 11.00 top(1) cumulative patch
and
PHKL_23901 11.00 remove stopped threads from load averages