how does HPbox report HW missing (V-class hpux11.0)

[Steve Post]

We have had some hardware problems. They got fixed fine. My question is on how HPUX reports these errors.

For the memory errors: The computer shutdown immediately, came back up without a portion of memory, and did NOT say anything about it. This was on a Sunday morning. (no email to root, nothing in syslog, no shutdown.log, no stm entry).

For disk errors: I ran vgdisplay -v vg00 and discovered one of the mirror disks was gone. No crash, but once again the computer did not tell me about it.

But if the tape is too dirty I get 2 emails about it.

Now I must say, the console gets messages but I don't like the idea of staring at 5 consoles 24x7, in the case an error shows up. Shouldn't I have gotten an email to root in these two cases???? I'm sure I could make little scripts to run ioscan every 30 seconds, but I would thing there is something standard in hpux11 that perhaps is missing????? I don't want to make some goofy script to fill a gap that I would expect to already be handled.

It's just a question.

[Rainer von Bongartz · Accepted solution]

Install and configure hardware monitoring thru EMS and you will
get every info you might want (even pager calls at 3 in the morning)

Reagards
Rainer

[Steve Post]

Thanks. It is already installed.
pause.....
Maybe I have something wrong with EMS? Maybe I just THINK it's installed? Is there a product id for it? Does it run via cronjob? Is there a log file that tracks the errors someplace? Is there an option to ignore disk, hardware, and memory errors (and it is switched on)?

[Christopher McCray_1]

Sounds to me like you do not have EMS/STM monitors set up to monitor these areas. Run /etc/opt/resmon/lbin/monconfig and select (C)heck monitors to see what is being monitored, and also make sure that your monitors areup at all. You can try doing a ps -ef|grep stm to see the stm processes are running.

Alternatively, you can bounce the diags by running /sbin/init.d/diagnostic stop, /sbin/init.d/diagnostic start

you can also run /etc/opt/resmon/lbin/set_fixed -l to see monitors status as UP or DOWN. do a /etc/opt/resmon/lbin/set_fixed -n \* to attempt to bring them up. If you have any other questions or anything else to add that can help me help you, please let me know.

Good luck
Chris

[James R. Ferguson]

Hi Steve:

You definitely need to install DIAGNOSTICS (EMS, STM and Predictive Support). You can get all three from the SupportPlus CDROM from the depot /cdrom/DIAGNOSTICS/B.11.00 when you mount at /cdrom.

Alternately, you can download these from the ITRC:

http://software.hp.com/cgi-bin/swdepot_parser.cgi/cgi/displayProductInfo.pl?productNumber=B6191AAE

An internal modem is often configured for use with Predictive Support and can be setup to only allow call-out. The setup of Predictive is menu-based (/opt/pred/bin/psconfig). You can configure it to run nightly; verify your hardware "health"; send logs to HP; mail notices of problems to yourself; AND an engineer will call you if there are issues!

Predictive documentation:

http://docs.hp.com/hpux/onlinedocs/H2571-90009/H2571-90009.html

STM (Support Tools Manager) is also extremely valuable. You can use it to see your installed memory, check your PDC version, etc.

For more information on the suite of diagnostic tools, see:

http://docs.hp.com/hpux/diag/index.html

Regards!

...JRF...

[Rainer von Bongartz]

Steve,

check your files /etc/rc.config.d/ems and /etc/rc.config.d/emsagtconf
In there you should set
EMS_ENABLED=1
and
AUTOSTART_EMSAGT=1

and then start ems
/sbin/init.d/ems start

check for processes /etc/opt/resmon/lbin/emsagent
/etc/opt/resmon/lbin/p_client

Reagards
Rainer

[Steve Post]

Next time I'm allowed shutdown the system, I'll reinstall a later version of diagnostics. I think that's where my error may be.

I have EMS and STM. But the EMS is from 1998.
I have the /etc/rc.config.d/config files set up ok.
I DON'T have /etc/opt/resmon/lbin/set_fixed or /monconfig.
But I have emsagent, p_client, and registrar in this directory. (date stamps from 1998).

and there's a client.log that says "Failed to connect to ,
IP address ###.###.###.###, Port 1712: Invalid argument.

Let me look at my OTHER V-class machine.....
The commands are there. It's been awhile. Maybe it DID report for memory error on this box and I forgot about it? Perhaps I am deluding myself into a false sense of security.
Yep. That's it.

Now I HAVE seen error emails from these two monitors. But not on memory dying, or disks disapearing. For box #1, I'll upgrade diagnostics. For box #2,...I guess I should start studying.

Thanks for the help. ^_^
Steve