One of processor is loaded much more then others

[Dmitriy_16]

I have added 2 more CPU to HP RP 5470 Server. Now system has 4 processors.
One of processors (number 3) is always loaded more then other three. Is it normal behaviour or there is some problem with configuration?
System: scp1 Fri Mar 4 16:04:11 2005
Load averages: 0.08, 0.05, 0.05
138 processes: 129 sleeping, 9 running
Cpu states:
CPU LOAD USER NICE SYS IDLE BLOCK SWAIT INTR SSYS
0 0.00 1.2% 0.0% 1.6% 97.2% 0.0% 0.0% 0.0% 0.0%
1 0.17 0.0% 0.0% 1.0% 99.0% 0.0% 0.0% 0.0% 0.0%
2 0.11 0.4% 0.0% 6.2% 93.4% 0.0% 0.0% 0.0% 0.0%
3 0.05 3.0% 0.0% 30.0% 67.0% 0.0% 0.0% 0.0% 0.0%
--- ---- ----- ----- ----- ----- ----- ----- ----- -----
avg 0.08 1.2% 0.0% 9.7% 89.1% 0.0% 0.0% 0.0% 0.0%

Memory: 1514084K (1468080K) real, 2067628K (2003136K) virtual, 1840636K free Page# 1/5

CPU TTY PID USERNAME PRI NI SIZE RES STATE TIME %WCPU %CPU COMMAND
3 ? 5766 scu 154 20 427M 110M sleep 334:36 1.76 1.76 scf
1 ? 5767 scu 154 20 427M 110M sleep 334:50 1.71 1.71 scf
2 ? 5768 scu 154 20 427M 110M sleep 336:04 1.65 1.65 scf
0 ? 1601 root 152 20 290M 87732K run 301:27 1.43 1.43 java
3 ? 5762 scu 154 20 35440K 25192K sleep 359:31 1.35 1.35 manager
2 ? 37 root 152 20 8064K 8064K run 93:39 1.01 1.00 vxfsd
3 ? 6528 root 158 20 2816K 508K sleep 238:11 0.63 0.63 ksh
1 ? 5765 scu 154 20 518M 241M sleep 152:10 0.62 0.62 scfserver
2 ? 5605 informix 180 20 284M 6320K sleep 94:52 0.52 0.52 oninit
0 ? 6233 fep 154 20 29128K 20872K sleep 169:33 0.52 0.52 fep
3 ? 5607 informix 180 20 284M 5672K sleep 85:21 0.32 0.32 oninit
2 ? 5769 scu 154 20 23896K 13556K sleep 138:20 0.30 0.30 oamagent
1 ? 5459 root 20 20 11116K 7712K sleep 88:28 0.27 0.27 cmcld
3 ? 5640 informix 180 20 284M 3996K sleep 59:52 0.26 0.26 oninit
0 ? 1103 root 152 20 15272K 3936K run 0:29 0.23 0.23 dmisp
2 ? 5764 scu 154 20 22260K 10068K sleep 67:21 0.23 0.22 smfagent
3 ? 5799 scu 158 20 2056K 424K sleep 56:23 0.17 0.17 sh
0 pts/ta 13971 scu 178 20 4680K 724K run 0:00 0.46 0.12 top
0 ? 6310 ospfep 154 20 11244K 1468K sleep 41:40 0.12 0.12 monfep
2 ? 0 root 127 20 32K 0K sleep 33:45 0.10 0.10 swapper
0 ? 1361 root 152 20 57180K 38748K run 2:48 0.10 0.10 vxsvc
0 ? 451 root 152 20 2172K 360K run 3:28 0.08 0.08 syncer
0 pts/ta 1264 root 154 20 3628K 288K sleep 0:00 0.05 0.05 telnetd
0 ? 1199 root 154 20 2652K 368K sleep 14:04 0.04 0.04 pwgrd
1 ? 5632 informix 180 20 284M 3516K sleep 1:16 0.04 0.04 oninit
3 ? 1114 root 152 20 13028K 2732K run 0:03 0.04 0.04 hpuxci
1 ? 1129 root 152 20 11912K 1796K run 0:03 0.04 0.04 swci
3 ? 3 root 128 20 32K 32K sleep 14:40 0.04 0.04 statdaemon
0 ? 5631 informix 180 20 284M 3536K sleep 1:02 0.04 0.04 oninit
0 ? 2519 root 154 10 4728K 1148K sleep 13:38 0.03 0.03 psmctd
1 ? 1 root 168 20 484K 204K sleep 10:10 0.02 0.02 init
3 ? 2 root 128 20 32K 32K sleep 0:03 0.02 0.02 vhand
1 ? 4 root 128 20 32K 32K sleep 0:41 0.02 0.02 unhashdaemon

[Simon Hargrave]

With such a low CPU load I wouldn't worry.

The output of top shows you which process is using which CPU. Looking at your output the processes are distributed across all CPUs.

Any thread has affinity to one CPU. So when you have such a low CPU load, one process that's slightly busier will load that CPU more, not affecting the others. For example your "manager" process is running on CPU 3
and using more CPU time than say, "scfserver" on CPU 1. It's just that the manager process is only using once CPU, and it happens to be slightly busier than scfserver.

[harry d brown jr]

Note vhand on processor 3 (well really 4). Do a search on vhand in this forum.

live free or die
harry d brown jr

[Jeff Schussele]

Hi Dmitriy,

This can be normal if you have heavy I/O on a specific driver like the fc fibre channel or igelan Gig-E driver.
System assigns a specific processor for each driver at boot time and if that driver is being utilized a lot then the CPU will reflect that.
We had a V-class with 8 FC cards & when under load it would drive the "selected" CPU to 100% & peg it there until the load dropped.

Rgds,
Jeff