diagmond takes excessive cpu usage

[jagdeesh]

This is 9000/800/N4000-75,running HP-UX 11i. whin i see on top, diagmond is taking nearly 12-13% of cpu usage. STM vesion is A.33.00. when i try to upgrade by installing PHSS_28251,it gives may install only on previously installed filesets. can any one help in this case?

Thanks in advance.
Jags

[Ken Hubnik_2]

Take a look at this patch.
PHSS_28251

[Ken Hubnik_2]

Sorry I meant to include the symptom description the patch fixes. Here it is

Symptoms:
PHSS_28251:
1. DTS JAGae50526
The STM diagnostic daemon, diagmond, uses excessive CPU
time.

[jagdeesh]

Thank you Ken, but i tried installing that patch. While installing its giving error "may install on previously installed base filesets" and execution phase getting fail. I seen the description too, there is no dependencies for it. How do i upgrade STM..??

[harry d brown jr]

You need to upgrade/reinstall STM first:

Other Dependencies:

This patch applies only to STM Sep 2002 SD-UX swlist bundles versioned under B.11.11.08.12 or STM version A.34.00, A.34.05, A.34.10, or A.34.15.


live free or die
harry

[harry d brown jr]

Try getting a newer STM from here:
http://www.software.hp.com/cgi-bin/swdepot_parser.cgi/cgi/displayProductInfo.pl?productNumber=B6191AAE

live free or die
harry

[Andrew Merritt_2]

This is a known problem which affects several releases of the OnlineDiags. It will be fixed in the March 2003 release (HWE0303).

In the meantime, there is a patch available for A.34.00 (September 2002 HWE), as Ken referred to, PHSS_28251.

Patches are in progress and should soon be available for A.31.00 (June HWE) and A.35.00 (December HWE):

PHSS_28250 HWE0206 A.31.00 11.11
PHSS_28252 HWE0212 A.35.00 11.11

Note that if you follow the link Harry posted, you will get the December release of OnlineDiag (A.35.00), which does not have the fix for JAGae50526. You would need to apply PHSS_28252 when it becomes available.

Andrew

[T G Manikandan]

check this patch

PHSS_27810

This should fix this

[Andrew Merritt_2 · Accepted solution]

PHSS_27810 is a patch for A.31.00, the June 2002 HWE, so won't help in the case where the current version is A.33.00 (the vPar release of OnlineDiag). A.33.00 has vPar support, and A.31.00 does not, so removing A.33.00 and going back to A.31.00 is not a good idea.

PHSS_27810 patch only partially fixes the diagmond CPU usage problem, fixing CR JAGae37841. This does deal with most of the diagmond performance problem, but JAGae50526 covers a remaining less serious CPU usage issue.

The best option from A.33.00 is to upgrade to A.34.00 (HWE0209) and install PHSS_28251.

The second option is to upgrade to A.35.00 (HWE0212) now, and install PHSS_28252 when it is available. A.35.00 includes the fix for JAGae37841, so the effect of the performance problem should be reduced significantly.

Andrew

[jagdeesh]

Hello.. Thank you all.
I upgraded the STM to A.34.00 and applied the patch 28251. I had to install PHSS_26947 since it was giving errors to some filesets. After upgradation and installing PHSS_28251, everything is normal and the STM revision si A.34.20. I have a vpar in this system. I want to follow the same procedur and upgrade on the other partition also.

Once again thanks to you all for your contribution in resolving this issue.

Best regards,
Jagdeesh.

[Darrell Tschakert]

Jagdeesh,
Where did you find the A.34.00 version of STM. We have A.31.00 and are having a problem similar to yours except that diagmond also leaks memory like crazy. I have gone to the software.hp.com site but all I can find is the A.35.00 version from Dec, 2002.

Thanks,
Darrell Tschakert
202/663-4436

[Andrew Merritt_2]

Darrell, if you have the June release (A.31.00) of OnlineDiags, you can apply patch PHSS_28250 (for 11.11).

PHSS_28250 fixes JAGae47777, which is a memory leak in diagmond related to OLAR support (the more OLAR slots, the bigger the leak). This does not affect HPUX 11.00 systems.

There should also be no problem installing A.35.00, as PHSS_28252 is now available. This includes the fixes for both JAGae47777 (diagmond memory leak) and JAGae50526 (excessive CPU usage by diagmond).