Lies and dammed lies.

[Paula J Frazer-Campbell]

Hi To All

Output from top: -

Load averages: 2.09, 1.46, 1.24
1128 processes: 1121 sleeping, 7 running
Cpu states:

CPU LOAD USER NICE SYS IDLE BLOCK SWAIT INTR SSYS
0 2.21 29.6% 39.9% 19.6% 10.9% 0.0% 0.0% 0.0% 0.0%
3 2.39 61.5% 0.8% 15.8% 22.0% 0.0% 0.0% 0.0% 0.0%
4 1.64 67.9% 1.2% 11.7% 19.3% 0.0% 0.0% 0.0% 0.0%
5 2.11 25.7% 33.5% 16.5% 24.3% 0.0% 0.0% 0.0% 0.0%
--- ---- ----- ----- ----- ----- ----- ----- ----- -----
avg 2.09 0.0% 0.0% 0.0% 100.0% 0.0% 0.0% 0.0% 0.0%

Note that the avg. row does not add up to what is going on the cpus.

This is not a momentary display but consistent.

Any Ideas ???


Paula

[Darrell Allen]

Hi Paula,

Patches perhaps? Top looks good on my servers.

Darrell

[Paula J Frazer-Campbell]

Hi Darrell

It could be patches but the strange thing is has just started telling me lies -

Yesterday fine today not - perhaps in on an early wind down for Xmas.

Baa Humbug.

Paula

[someone_4]

hmm
here is my out put from 2 servers and they both seem fine

Cpu states:
CPU LOAD USER NICE SYS IDLE BLOCK SWAIT INTR SSYS
0 0.23 0.0% 0.0% 1.2% 98.8% 0.0% 0.0% 0.0% 0.0%
1 0.85 61.7% 0.0% 0.6% 37.7% 0.0% 0.0% 0.0% 0.0%
2 0.80 45.7% 0.0% 0.6% 53.7% 0.0% 0.0% 0.0% 0.0%
3 0.23 42.5% 0.0% 0.4% 57.1% 0.0% 0.0% 0.0% 0.0%
4 0.09 49.7% 0.0% 0.4% 49.9% 0.0% 0.0% 0.0% 0.0%
5 1.28 22.2% 0.0% 1.2% 76.6% 0.0% 0.0% 0.0% 0.0%
6 0.60 8.0% 0.0% 2.2% 89.8% 0.0% 0.0% 0.0% 0.0%
7 0.20 6.4% 0.0% 1.8% 91.8% 0.0% 0.0% 0.0% 0.0%
--- ---- ----- ----- ----- ----- ----- ----- ----- -----
avg 0.54 29.6% 0.0% 1.0% 69.4% 0.0% 0.0% 0.0% 0.0%

CPU LOAD USER NICE SYS IDLE BLOCK SWAIT INTR SSYS
0 4.08 30.7% 0.0% 69.3% 0.0% 0.0% 0.0% 0.0% 0.0%
1 5.42 23.8% 0.0% 76.2% 0.0% 0.0% 0.0% 0.0% 0.0%
--- ---- ----- ----- ----- ----- ----- ----- ----- -----
avg 4.75 27.7% 0.0% 72.3% 0.0% 0.0% 0.0% 0.0% 0.0%

[Darrell Allen · Accepted solution]

Should have included:
PHCO_22686 on my 11.0 servers.

Darrell

[Roger Baptiste]

paula,

Top is not a reliable diagnostic tool. It is good for a quick snapshot on what is going on the system. I have run in incorrect process numbers in top and have since then used glance. I believe there are patches for TOP problems. Check it out.

Have a look at this document:
http://us-support.external.hp.com/cki/bin/doc.pl/sid=1a215576091588638f/screen=ckiDisplayDocument?docId=200000026600844

BTW, how many cpu this system has? it doesn''t seem to show all the cpu''s in the top output.

HTH
raj

[Roger Baptiste]

Hi paula,

Here is the link to fix(hopefully) your problem:

http://us-support.external.hp.com/cki/bin/doc.pl/sid=0399507f10ca641e08/screen=ckiDisplayDocument?docId=200000024631586

I see some unusual statistics on a K400 system I am using as
a production system. The output is similar to this:

Load averages: 0.00, 0.00, 0.00
1231 processes: 1144 sleeping, 15 waiting, 1 starting
Cpu states:
CPU LOAD USER NICE SYS IDLE BLOCK SWAIT INTR SSYS
0 0.00 -13.4% -11.4% 36.8% 5.5% 4.3% 22.4% 4.8% 50.9%
1 0.00 25.7% -50.2% 7.3% 47.5% 98.9% -16.1% 71.6% -84.7%
2 0.00 -5.2% -5.1% 29.5% 49.0% 14.1% 37.6% 8.6% -28.6%
3 0.00 31.4% -57.5% 8.2% 43.1% 50.8% 68.7% 22.0% -66.7%
4 0.00 11.0% -52.2% 9.6% 109.5% -20.2% 27.9% 21.2% -6.8%
--- ---- ----- ----- ----- ----- ----- ----- ----- -----
avg 0.00 301.4% -1089.0% -1833.0% -196.8% 1716.4% 478.8% 575.1% 147.2%

How can the LOAD average be 0.00, considering the values in the remaining
column?

Configuration Info

Operating System - HP-UX
Version -10.01
Hardware System -K400

Solution

There is a problem with the output of the top command at 10.0. When
the system is lightly loaded, everything is okay, but when system
load increases, as in your case, the output becomes very strange.

The problem is caused by top not performing its bounds checking and
overwriting wrong areas of memory.

HP-UX 10.10 contains a fix for this problem, while a patch is available
for 10.0 and 10.01. The patch, found on XPATCH under Supplemental Patches,
is rc_patches/WTEC/TOP.

Refer to the top(1) man page for details o

[Paula J Frazer-Campbell]

Hi

Well spotted - system qty 6 CPUs of which qty 2 are ICOD and do not display in top - which is normal.

Paula

[Paula J Frazer-Campbell]

More info:-

Server is N Class running 11.00.

Paula

[someone_4]

Does any one know how we can tell what version of top we are running?

[Darrell Allen]

Baa Humbug? We haven't even had Thanksgiving yet. But I guess that's an American holiday that the rest of the world misses out on. Pity!

Of course the stores have been full of Christmas merchandise since Halloween so I guess the Christmas season will soon be year round anyway.

Happy (you pick it) Holiday!

Darrell

[Darrell Allen]

Richard, you ask a good question. I'd like to know an easy way. I happened to know I had a patch for top so I did:

swlist -l fileset | grep -i top
which top
what /usr/bin/top

Then I went to the ITRC and verified my what output matched that listed in the description for my patch.

There's gotta be a better way.

Darrell

[Roger Baptiste]

Paula,

Do you have this patch?
PHCO_17439 1.0 top(1) cumulative patch
(this is for Nbox on 11.00)

also, check the top details:
***
what /usr/bin/top
/usr/bin/top:
top.c $Revision: 82.1.1.6 $ $Date: 99/01/20 04:07:31 $
find_ttys.c $Revision: 80.1 $ $Date: 96/09/06 17:19:38 $
PATCH_11_00: top.o find_ttys.o 99/01/21
>ll /usr/bin/top
-r-xr-sr-x 1 bin sys 32768 Jan 21 1999 /usr/bin/top
****

HTH
raj

[Paula J Frazer-Campbell]

Hi
root@N-0/>what /usr/bin/top
/usr/bin/top:
top.c $Revision: 82.1.1.15 $ $Date: 2000/07/19 15:29:13 $
find_ttys.c $Revision: 80.1 $ $Date: 96/09/06 17:19:38 $
PATCH_11_00: top.o find_ttys.o 00/07/20

root@N-0/>ll /usr/bin/top
-r-xr-sr-x 1 bin sys 32768 Jul 20 2000 /usr/bin/top


Paula

[Paula J Frazer-Campbell]

Hi

It looks like a patch problem - Darrell pointed me in this direction.

PHCO_21892
In a multi processor system, if any processor other than the last processor is deallocated, #
top(1) shows incorrect cpu usage, average cpu usage and load averages information.


Paula

[Paula J Frazer-Campbell]

BTW

The First Xmas decorations I saw in London were on the 9th October - Daft or what.



Paula

[someone_4]

Here is what I have

/# what /usr/bin/top
/usr/bin/top:
top.c $Revision: 82.1.1.15 $ $Date: 2000/07/19 15:29:13 $
find_ttys.c $Revision: 80.1 $ $Date: 96/09/06 17:19:38 $
PATCH_11_00: top.o find_ttys.o 00/07/20
/# ll /usr/bin/top
-r-xr-sr-x 1 bin sys 32768 Jul 20 2000 /usr/bin/top
/#

[harry d brown jr]

That's why I use glance/measureware. Top, sar, vmstat, iostat, uptime are good for a quick look, but they do not have the kernel hooks that glance/measureware does, and in a risc machine where a process can execute multiple commands in a single cycle, top don't be able to see it happen.


live free or die
harry

[Roger Baptiste]

Paula,

I have 82.1.1.6 version on all systems, even the ones which were patched recently.

Do you have the cumulative top patch??-> PHCO_17439

HTH
raj

[Paula J Frazer-Campbell]

Raj

Yep got it:-

# PHCO_17439 1.0 top(1) cumulative patch
# PHCO_21892 1.0 top(1) cumulative patch
# PHKL_20170 1.0 Dino (GSCtoPCI) spinlock


Paula

[Darrell Allen]

And PHCO_22686 supersedes those top patches.

What output:

OS-Core.CMDS-AUX,fr=B.11.00,fa=HP-UX_B.11.00_32/64,v=HP:
/usr/bin/top:
top.c $Revision: 82.1.1.17 $ $Date: 2000/12/02 08:0
1:21 $
find_ttys.c $Revision: 80.1 $ $Date: 96/09/06 17:19
:38 $
PATCH_11_00: top.o find_ttys.o 00/12/03

I hate this small window! I never know what cut and paste is going to look like until after posting.

Darrell

[Paula J Frazer-Campbell]

Hi

Thanks to all

A Patching we will go :- Traa Laa

ps :-

Remember to sing the:-

5 Gold Rings bit as loud as possible at Xmas.

XXX

Paula

[Paula J Frazer-Campbell]

Hi
For information only:-

Server had been up for circa 120 days and top is run 24/7, 6 processors - qty 2 ICOD.

I had made no changes to the server - one day top OK next not.

Top had run into a counter limit which affected the average row.

If this occurs then this top works

root@N-0/usr/bin>what top
top:
top.c $Revision: 82.1.1.17 $ $Date: 2000/12/02 08:01:21 $
find_ttys.c $Revision: 80.1 $ $Date: 96/09/06 17:19:38 $

Thanks to Phil Rassell at HP.

Paula

[T. M. Louah]

To add to above, you run checksum on top compare it with an existing system top.
.. r u running 11i ? or vPar?
tx

[Paula J Frazer-Campbell]

Hi

The problem is fixed I was just passing information on in case anyone else should have the same problem.

Paula