diagmond

[Scott D. Allen]

I've got an HP9000 10.20 system running and I'm monitoring performance (top/glance). I see that the diagmond has a pretty significant %CPU. Is this normal. I'm assuming that this is just the diagnostics daemon running in the background collecting statistics. It's ~9%CPU Time.

--Scott

[Scott D. Allen]

Rick,
Sorry. I didn't understand the question. I'm not familiar with the stm daemon.

--Scott

[Rick Garland]

The diagmond is the stm daemon.

There are all kinds of files, including logs, under the /var/stm directory area.

Would there be another process running by chance?

[Rick Garland]

diagmond keeps track of the system HW config which includes a list of tools available for each device. Also keeps track of which tools have been executed or are executing on each device. (From the man diagmond)

If you were to go into stm (xstm, cstm, or mstm) are you seeing a lot of activity occurring? You might see them in your syslog file as well. If you are not running stm, could someone else be?

The CPU usage can vary for this process. Depends on what is happening behind the scenes.

[rene martinez · Accepted solution]

You need patch PHSS_20007. The problem you are experiencing with giagmond is a known problem. This patch will correct it.

[Vincente Fernandes]

stm uses 2 daemons. "dialogd" & "diagmond". stm is part of OnlineDiag B.10.20.11.08 HPUX 10.0 Support Tools Bundle
Check if you have any cron job for gathering stm information.

[Scott D. Allen]

Renee, How do I get information on PHSS_20007? Also, I can't seem to get access to download that patch individually. Do I have to create a custom patch bundle or something? Is this under Maint./Support?

Others, Thanks for the info on STM. No, we don't have any STM users or jobs running. Does this have to be enabled if noone is using STM? Is this required for glance/top? Is there any adverse affect to disabling this (DIAGNOSTICS=0 in /etc/rc.config.d/diagnostic)?

--Scott

[melvyn burnard]

PHSS_20007 is for HP-UX 11.x
For 10.20 you need:
S800 systems PHSS_20006
S700 systems PHSS_20005

HTH

[Devbinder Singh Marway]

Hi,

If you have predict installed (i.e. HP dial in and check out system for errors i.e. h/w errors , disk firmware upgrades etc..), you need the diagmond daemon running ( it will say so when you run /opt/pred/bin/psconfig )

if not then you can stop diagnostics editing the files you mentioned on reboot or do
/sbin/init.d/diagnostic stop

regards
Dev

[Scott D. Allen]

I guess one of the most notable problems related to this diagmond issue is the fact that I'm showing excessively high swap usage, though I can't figure out why and it doesn't show any of my swap disks in use....
$ swapinfo -t
Kb Kb Kb PCT START/ Kb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 917504 0 917504 0% 0 - 1 /dev/vg00/lvol2
dev 819200 0 819200 0% 0 - 2 /dev/vg06/lvol2
reserve - 1474296 -1474296
total 1736704 1474296 262408 85% - 0 -
$

Suggestions?

--Scott

[John Palmer]

There is a recent problem on 11.00 where one of the diagnostic daemons has a memory leak. I'm not sure if this is a 10.20 problem also but it might be worth shutting and restarting diagnostics to see if your used swap space reduces dramatically.

/sbin/init.d/diagnostic stop
/sbin/init.d/diagnostic start

Regards,

John

[Scott D. Allen]

Is that what patch PHSS_20005/6/7 is supposed to fix, getting back to the original thread? And, what's the meaning of that swapinfo command which shows 85% total used, but 0% of both swap devices used?

--Scott

[Anthony deRito]

Scott, 85% total used includes the virtual memory "reserved" for processes that are currently running. It does not necessarioy mean that it is all consumed by pageout activities. The swap usage column is a parameter for the percentage of swap used for paging out activities. This shows your not really doing any paging out which is a good thing.

Tony

[John Palmer]

The system has reserved about 1.4Gb of swap space in case it needs to use it. This will be the sum total of data segment space used by processes. If this is more than usual then it could be due to a memory leak in a process. I found a similar problem last week in one of the diagnostic daemons where it had grown its data space to > 600Mb. Killing it and restarting reduced reserved swap from over 85% to 40%.

From the output of swapinfo it looks as though you haven't got swapmem_on set to 1 in your kernel. It would be advisable to set this when you get the chance as it reduces your overall device swap requirements by 75% of the memory in your server which I guess must be >= 2Gb.

Regards,

John

[Scott D. Allen]

Nice! I've finally got it. Last question: Is there a way to tell what processes are making up the "reserve" portion, or is it just a guestimate and/or based on the number of logins? We recently changed large_ncargs_enabled to 1.....giving a 2MB arg space. Could this be why the large "reserve" values are coming up?

Thanks to all.

--Scott

[Rick Garland]

What is taking up the reserve is the OS itself.

[Anthony deRito]

To execute a process, the kernel creates a per-process virtual address space that is set up by the kernel. Portions of the virtual space are mapped onto physical memory but not all of it. Most of it can be kept out in VM. If the kernel needs it, it will get it by doing a translation lookup.

The kernel sets up enough reserved memory in order to be able to swap-out the entire process if there is too much memory pressure on the system. You have no control over this. This figure is pre-calculated by the kernel. This is why you somtimes see the message:

"Out of memory"

When trying to compile code. The kernel wants to "reserve" memory "just-in-case" it needs it.


Tony

[CHRIS_ANORUO]

Hi Scott,
TO add to the previous answers, The Reserve is the Paging space on reserve. This is the amount of paging space that could be needed by processes that are currently running, but that has not yet been allocated from one of the paging areas. When a process is created , or requests additional space, space isreserved for it by increasing space shown on the reserve line above. Paging areas are enabled at boot time on by the swapon command.
If you want to see processes on you server, you can us ps command or lsof( download and install the software from this link: http://hpux.cs.utah.edu/hppd/hpux/Sysadmin/lsof-4.48/) It is a Hp-UX Porting site.

Cheers!

Chris