check_esxi_hardware + ESXi 5.0 + DELL OMSA = FAIL

Written by - 19 comments

Published on November 8th 2011 - Listed in Nagios Hardware VMware Virtualization Monitoring


Attention users of the Nagios plugin check_esxi_hardware !
If you want to use the plugin on a DELL server with ESXi 5.0 installed, you have to make sure that OMSA is NOT installed. Otherwise the plugin won't work.

For some strange reason, there seems to be a problem in this package (Dell_OpenManage_ESXi_OM650-offline_bundle-467660.zip) and also in the ESXi 5.0 version which comes bundles with OMSA.

Another point to look at is the VMware ESXi tools compatibility guide:
http://www.vmware.com/resources/compatibility/vcl/esxi.php
DELL clearly seems to be missing in that list, while other CIM providers (e.g. from HP) are shown.

Thanks to Dave Stevens and Pascal Saul for their feedbacks and tests!

Update 13.11.2011: Take a look at this page, which officially quotes Dell, that the current version is not fully ESXi 5.0 compatible:
http://www.delltechcenter.com/page/OpenManage+Support+for+VMware+ESX+5.0

Update 19.12.2011: A new version of the Offline Bundle has been released and was successfully tested. The plugin now works again. Please take a look at the following blog post for more information: Dell OMSA + ESXi 5.0 is now working!


Add a comment

Show form to leave a comment

Comments (newest first)

romain0412 from wrote on Dec 15th, 2011:

Thanks for your help :-)

Romain


Claudio from Switzerland wrote on Dec 15th, 2011:

Since ESXi 4.1 it is not possible anymore to use a read-only user for this purpose.
However you can either hide the password by using a password file (see documentation) or you can add another user in vsphere client by following these steps: http://www.claudiokuenzler.com/blog/114/check-esxi-wbem-esxi-4.1-user-authorization


romain meunier from wrote on Dec 15th, 2011:

Sorry,

I have forgotten the firewall. I have an another question, what permission must have the user?

A read-only user can be used with this script?

Thanks

Romain


Claudio from Switzerland wrote on Dec 15th, 2011:

The connection to port 5989 is necessary to retrieve the CIM information. Is there a firewall between Nagios server and ESXi server?
Also, as said before, check if you have installed the Offline Bundle by using the vSphere Perl API:
vihostupdate.pl --server my_ip --query
If the HP Offline Bundle isn\'t listed then you need to install it.


romain meunier from wrote on Dec 15th, 2011:

Claudio,

I can't establish connection on port 5989...

That's seems that I must configured something on my ESXi?

I can see hardware status directly on the ESXi with my vSphere client, that seems that CIM is activated, isn't it?

Thanks


Claudio from Switzerland wrote on Dec 15th, 2011:

Sorry, I meant 5989, not 5589.
Check if you can establish a connection on that port. If that works, double-check if you have installed the HP CIM Offline Bundle for ESXi 4.1 and that CIM settings are activated in the configuration (in vSphere client select the ESXi server, Configuration tab, Software table click on Advanced Settings, enable everything mentioning CIM).


romain0412 from wrote on Dec 15th, 2011:

I can't connect to port 5589 by telnet, but I can on 443.


Claudio from Switzerland wrote on Dec 15th, 2011:

It seems that you don't have the HP MIB/CIM Offline Bundle installed. Can you double-check that?
Also can you open a connection from Nagios server to your ESX/i server on ports 5589 and 443 ?


romain0412 from wrote on Dec 15th, 2011:

Claudio,
I have tried to delete the fisrt entry, the issue still persist. I have deleted the second entry, tested, the third, tested etc. And the issue still persist.

The problem may come from my ESXi, must I see something of the ESXi?

Thanks,


Claudio from Switzerland wrote on Dec 15th, 2011:

OK, open the plugin file in an editor like vi and delete the first entry in the ClassesToCheck array: OMC_SMASHFirmwareIdentity. It should look like this then:

ClassesToCheck = [
'CIM_Chassis',
'CIM_Card',
'CIM_ComputerSystem',
'CIM_NumericSensor',
'CIM_Memory',
'CIM_Processor',
'CIM_RecordLog',
'OMC_DiscreteSensor',
'OMC_Fan',
'OMC_PowerSupply',
'VMware_StorageExtent',
'VMware_Controller',
'VMware_StorageVolume',
'VMware_Battery',
'VMware_SASSATAPort'
]

Then try it again and let me know if the problem still exists.


romain0412 from wrote on Dec 15th, 2011:

The ESXi are installed on HP Proliant, I have try to add -V hp but same result.

Thanks,

Romain


Claudio from Switzerland wrote on Dec 15th, 2011:

Hi Romain,
On what hardware is the server running? Did you try to define the hardware type by using the -V parameter?


romain0412 from France wrote on Dec 15th, 2011:

Hello,

I try to use your script in order to monitor ESXi 4.1 but when I launch the script I have this resuslt :

./check_esxi_hardware.py -H my_ip -U my_user -P my_passwd -t 20 -v
20111215 16:03:58 Connection to https://my_ip
20111215 16:03:58 Check classe OMC_SMASHFirmwareIdentity
CRITICAL: Execution time too long!

I don't find the issue reasin. Have you got an idea?

Thanks,

Romain


Claudio from Switzerland wrote on Nov 19th, 2011:

check_esx3 checks your ESXi physical host (CPU, Memory, ...) but you can also check the virtual guests. I do not recommend to do so though (due to intensive resource usage). You should check your virtual guests like any other normal host. E.g. nrpe for Linux and NSClient++ for Windows guests.


O'Shaughnessy Evans from United States wrote on Nov 18th, 2011:

Thanks. I was looking at that plugin. I appreciate the tip about resource usage -- definitely a factor I need to be careful with. Nagios/Cacti boxes seem to get pretty busy! So just to make sure I understand correctly, check_esx3 would report on guest resources, not on the ESXi layer itself, right?


Claudio from Switzerland wrote on Nov 18th, 2011:

Well I'm not an expert in CIM but even if it were possible to monitor things like CPU load, filesystems etc via CIM, I would still strongly suggest to do this directly on the target OS. Personally I use OP5's plugin check_esx3 which uses the vSphere Perl API to check exactly those things (CPU load, memory usage and vmfs usage). You should give it a try. But watch out, it takes a lot of resources on your Nagios server. I set up a check once every 30min to decrease server load.


O'Shaughnessy Evans from United States wrote on Nov 18th, 2011:

Okay, I think I'm starting to understand this better. Am I correct in thinking that there's little value in monitoring OS resources (filesystems, free memory, CPU load) at the ESXi layer, between the guests and the hardware, because that's never really the actual problem point? The real problem is always going to be at the guest OS layer or at the hardware layer, but never between because of the way that whole stack is structured and utilized?


Claudio from Switzerland wrote on Nov 18th, 2011:

Nope, that's not possible, unless the hardware vendor added this into a special, non-standard CIM table. The available CIM classes defined by VMware can be found here and contain only hardware related information.


O'Shaughnessy Evans from United States wrote on Nov 18th, 2011:

Hi. I was wondering if the plugin checks for any alerts at the OS layer, like a spiked CPU, lack of memory, or a full filesystem. I am new to ESXi, and it is hard to tell from the CIM API if that kind of information is even available.

(BTW, I am using the plugin against Dell C6100 blades, and it seems to be working fine.)

Thank you.