disable EMS HW monitor polling on specific devices

[Godfrey Zhang]

I need to disable SCSI_tape monitoring/polling, so I did monconfig as follows:

The current monitor configuration is:
1) Send events generated by monitors
/storage/events/disks/default
/adapters/events/FC_adapter
/system/events/core_hw
/system/events/memory
/system/events/cpu/lpmc
/system/events/system_status
with severity >= MAJOR WARNING to EMAIL root


But when I checked api.log, it did poll on SCSI_tape:

-------------------Start Event--------------------
User event occurred at Thu Mar 18 20:55:13.972289 2004
Process ID: 2999 (/usr/sbin/stm/uut/bin/tools/.../dm_stape) Log Level: Error
dm_stape monitor has successfully registered with diaglogd for hardware event monitoring.
-------------------End Event----------------------

And I can see the processes:
WLG1HXO6 (11iprod)# ps -ef | grep monitor
root 2755 1 0 20:55:05 ? 0:00 /usr/sbin/stm/uut/bin/tools/monitor/disk_em
root 2913 1 0 20:55:10 ? 0:04 /usr/sbin/stm/uut/bin/tools/monitor/dm_core_hw
root 2829 1 0 20:55:08 ? 0:03 /usr/sbin/stm/uut/bin/tools/monitor/dm_TL_adapter
root 2793 1 0 20:55:06 ? 0:00 /usr/sbin/stm/uut/bin/tools/monitor/dm_FCMS_adapter
root 3109 1 0 20:55:30 ? 0:00 /usr/sbin/stm/uut/bin/tools/monitor/ha_disk_array
root 3150 1 0 20:55:31 ? 0:00 /usr/sbin/stm/uut/bin/tools/monitor/sysstat_em
root 3121 1 0 20:55:30 ? 0:07 /usr/sbin/stm/uut/bin/tools/monitor/lpmc_em
root 2999 1 0 20:55:12 ? 0:03 /usr/sbin/stm/uut/bin/tools/monitor/dm_stape
root 2969 1 0 20:55:11 ? 0:00 /usr/sbin/stm/uut/bin/tools/monitor/dm_memory
root 22883 7325 1 12:18:03 pts/ta 0:00 grep monitor
WLG1HXO6 (11iprod)#

Even I restared EMS using monconfig (K and E).

Anybody can tell me how to disable it?

I expect no dm_stape process, and no persistance file for this monitor.

Thanks

[Nick Zraika]

On HP-UX 11.11 i have edited the following file to disable EMS notification for specific disks. I believe you can also enter in the hardware path of your tape unit..

check it out, filename:

/var/stm/data/tools/monitor/disabled_instances

p.s
You will need to restart ems

/sbin/init.d/ems stop
/sbin/init.d/ems start

goodluck

[Godfrey Zhang]

Nick,

The disabled_instances only disables the notification, but the hw monitor still polls the devices, I want to stop hw monitor polling every 2minutes on these devices.

[Hemanth Gurunath Basrur · Accepted solution]

Hello,

Please refer to EMS HW Monitors FAQ.

http://docs.hp.com/hpux/onlinedocs/diag/ems/ems_faq.htm

I have picked the following info from EMS HW Monitors user's guide available at http://docs.hp.com/hpux/onlinedocs/2512/ems.pdf

"You can disable hardware event monitoring if desired. However, all EMS Hardware Monitors will be disabled. You cannot disable a specific monitor. While monitoring is disabled, all monitoring requests are
disabled. The monitoring requests are retained and become operational when monitoring is re-enabled."

CAUTION:
Use careful consideration before disabling hardware event monitoring. Be aware that ALL
hardware monitoring will be disabled. While monitoring is disabled, your hardware resources are vulnerable to undetected failures. Disabling monitoring will impact MC/ServiceGuard if package dependencies have been created for the hardware event monitors.

To disable hardware event monitoring:
1. Run the Hardware Monitoring Request Manager by typing:
/etc/opt/resmon/lbin/monconfig
2. From the main menu selection prompt, enter K.
3. Confirm disabling when prompted to do so. When you are ready to re-enable hardware event monitoring,see â Enabling Hardware Event Monitoringâ .

"Hardware monitoring can be disabled using the Hardware Monitoring Request Manager. Disabling monitoring disables all EMS Hardware Monitors. Individual monitors cannot be disabled using the Hardware
Monitoring Request Manager. When monitoring is disabled all existing monitoring requests are unregistered, and then a kill -2 command
is issued to stop all monitors."

HTH.

Regards,
H

[Godfrey Zhang]

Hi Hemanth, Nick;

Thanks for your help. I was hoping EMS HW Monitor working as usual but only some devices not checked. For example, I dont' want dm_stape checked, but rest all checked, thus I was hoping /usr/sbin/stm/uut/bin/tools/monitor/dm_stape wasn't running; and in /etc/opt/resmon/persistence/ no stape there.

If I delete stape in ./persistence/ and ./dictionary/, can this stop polling to stape?

[Nick Zraika]

I wouldn't delete them (as it may impact the manner in which ems behaves) to test i might re-name them, if it causes any issues simply re-name them back.... probably not the best way to do it but hey, if it works.. who's complaining :)

[Andrew Merritt_2]

Ok, first of all, what revision of OnlineDiags do you have installed, and secondly, how do you know it is actually polling the devices? What problem are you trying to solve?

In recent revisions of OnlineDiags (from memory, for the last two years), dm_stape by default does not poll the device, it only reports events logged by the drives. The default Poll frequency is zero, which means 'don't poll'. The dm_stape monitor will run, but not actively contact the devices.

Andrew

[Andrew Merritt_2]

You say:
> I want to stop hw monitor polling every 2minutes on these devices.

What are you seeing being logged every two minutes that suggests this?

If you're seeing

-------------------Start Event--------------------
User event occurred at Thu Mar 18 20:55:13.972289 2004
Process ID: 2999 (/usr/sbin/stm/uut/bin/tools/.../dm_stape) Log Level: Error
dm_stape monitor has successfully registered with diaglogd for hardware event monitoring.
-------------------End Event----------------------

every two minutes, that would indicate the monitor was dying and being restarted, which is a different problem. But the 'ps' listing would suggest otherwise, as the start time is the same for all the monitors.

Andrew

[Godfrey Zhang]

Thanks heaps everyone.

Andrew was right, my OnLineDiag version is Sep 2003, and the polling interval is set to zero. My assumption it was polling was the api.log popping with stape, so I thought it was polling the devices. The reason we wanted to stop polling was our SuperDLT stacker (6 drives) connected to a MDR, recently those tape drives randomly reset themselves causing backup failures, so we wanted to stop any unnecessary access to them to check where the problems are.

Thanks, Andrew. Cheers.



OnlineDiag B.11.00.26.07 HPUX 11.00 Support Tools Bundle, Sep 2003


POLL_INTERVAL 0 # polling interval in minutes

-------------------Start Event--------------------
User event occurred at Thu Mar 18 20:55:13.972289 2004
Process ID: 2999 (/usr/sbin/stm/uut/bin/tools/.../dm_stape) Log Level: Error
dm_stape monitor has successfully registered with diaglogd for hardware event monitoring.
-------------------End Event----------------------

-------------------Start Event--------------------
User event occurred at Fri Mar 19 14:47:59.545166 2004
Process ID: 2999 (/usr/sbin/stm/uut/bin/tools/.../dm_stape) Log Level: Error
Tool is exiting due to receipt of a SIGINT signal.
-------------------End Event----------------------


zhanggod@wlg1hxo6:: ps -ef | grep -i monitor
root 4502 4301 0 Mar 19 ? 0:20 /usr/sbin/stm/uut/bin/tools/monitor/dm_core_hw
root 4358 1 0 Mar 19 ? 0:00 /usr/sbin/stm/uut/bin/tools/monitor/dm_FCMS_adapter
root 4314 1 0 Mar 19 ? 0:00 /usr/sbin/stm/uut/bin/tools/monitor/disk_em
root 4581 4301 0 Mar 19 ? 0:00 /usr/sbin/stm/uut/bin/tools/monitor/dm_memory
root 4645 4301 0 Mar 19 ? 0:06 /usr/sbin/stm/uut/bin/tools/monitor/dm_stape
root 14440 14438 0 13:25:02 ? 0:00 /usr/bin/sh /var/opt/OV/bin/OpC/hpux/monitor/cpu_mon.sh hpux
root 4839 4301 0 Mar 19 ? 0:00 /usr/sbin/stm/uut/bin/tools/monitor/ha_disk_array
root 4875 4301 0 Mar 19 ? 0:01 /usr/sbin/stm/uut/bin/tools/monitor/sysstat_em
zhanggod 14455 14420 0 13:25:09 pts/ta 0:00 grep -i monitor
root 4391 4301 0 Mar 19 ? 0:02 /usr/sbin/stm/uut/bin/tools/monitor/dm_TL_adapter
root 4851 4301 0 Mar 19 ? 0:30 /usr/sbin/stm/uut/bin/tools/monitor/lpmc_em
zhanggod@wlg1hxo6::

[Andrew Merritt_2]

> The reason we wanted to stop polling was our
> SuperDLT stacker (6 drives) connected to a
> MDR, recently those tape drives randomly
> reset themselves causing backup failures, so
> we wanted to stop any unnecessary access to
> them to check where the problems are.

Did anything get changed or installed around the time you started noticing the problem?

I have seen a similar problem where HAO was running, as it runs the STM info tools, and running the tape info tool in a SAN environment when there is a backup running can cause problems.

If the problem persists, you could try disabling the monitor while the backup is running. If that clears the problem, then you should definitely open a support call with HP since that should no longer be happening with the polling being turned off.

(I've attached a document I found on ITRC that shows how to do this.) Note that I'm not recommending disabling dm_stape as a matter of course, only as an aid to troubleshooting this problem.

Andrew

[Godfrey Zhang]

Andrew,

Your last reply is exactly what I wanted. Thanks heaps. You deserve all the points, mate.