missing EMS catagories

[Rick Garland]

Hi all:

Working on rx4640 ia64 system running 11.23.

Working on setup of EMS. I am finding that there are multiple subcatagories missing, most notably the availMB from the /system/filesystem catagory.

Why would some subcatagories be missing?

[Kent Ostby]

There are multiple EMS monitors and I think they are packaged differently.

Might want to check what you have installed on this box versus another of your boxes.

[Rick Garland]

Got the same version of EMS on all of our ia64 systems.

Just this one system missing some catagories and subcatagories.

Another example, the 'NumUsers' is missing as well.

[Rick Garland]

New development:

A system I was working on (ia64, HPUX 11.23) is now without some of the catagories and subcatagories in the Resource Hierarchy.

Example, I was doing /system/filesystem/availMB and when I changed the trap destination in the snmpd.conf file, stop and restarted the /sbin/init.d/SnmpMaster it does not work anymore. I changed the trap dest back to the original value and bounced the SnmpMaster but to no avail.

Something to do with the snmp trap and the SnmpMaster (I believe). Still working it. Anybody else have this problem?

[Rick Garland]

Changing the /etc/snmpd.conf file back to the original and bouncing the SnmpMaster did not help. Tries rebooting system as well - no joy.

What happened? Any ideas?

[Rick Garland]

Still working this. Going through command line to get more descriptive messages.

# emscli -a /system/filesystem/availMB
Adding request ...
Error : /system/filesystem/availMB is not a valid resource instance.
Error : failed to register request.
#

This is what is happening when I try to add EMS request for filesystem monitoring. Have tried calling HP. Put on the patch they suggested but still no luck.

Anybody run into this before?

[Rick Garland]

Found a pattern I can communicate;

I modified the snmpd.conf file to input values for the get/set community names and for the trap dest. I then cycled the /sbin/init.d/SnmpMaster. Now I am having the trouble. I put the original snmpd.conf file back and cycled the SnmpMaster. Still having the trouble.

On another system (11.23, rx4640 - same HW, OS, etc...) I have not modified the snmpd.conf file or cycled the SnmpMaster. I get all of the resource instances from EMS.

It would seem that somehow the snmp process is hindering the EMS.

I am now afraid of cycling the SnmpMaster until I find out what is up.

Does anybody out there have an idea?

[Rick Garland]

After some research and several support calls, have finally found the answer.

The get-community-name in the snmpd.conf file was changed. As a result you need to change the invocation line for the EMS monitors to have the -c option.

There are 2 docs to review.
KBRC00012590 & UXEMSKBRC00003120

The bottom line is that because the default get-community-name was changed then need to tell EMS that it was changed and what the new value is.