Monitoring a server running a VMware vSphere ESXi v5 hypervisor

Note: This post was prepared by a technical expert, in the process of finding the best solution on a test server.

There was a need to organize monitoring of the correct operation of the LSI MegaRAID family controllers on servers running the VMware vSphere ESXi v5.5 hypervisor – and, accordingly, to automatically receive notifications in case of any failures (for example, failure of one of the HDDs). In the process of the testing, it turned out that the solution is not limited to hypervisor data stores.

To solve this task, we used a test server based on the Supermicro X9DR3/i-F motherboard with an LSI MegaRAID SAS 9260-4i controller, to which two HDDs were connected and RAID1 was configured.

Even though LSI MegaRAID SAS 9260-4i is officially supported by ESXi, you won't be able to get any RAID status information from the Health Status section of the VMware vSphere client:

Fortunately, this is fixable. We go to the lsi.com website and find the archive with «SMIS Provider» for the desired controller there:

Download, unpack and find the file with the «vib» extension. This is a package that provides controller status monitoring using the built-in ESXi sensor mechanism. Copy this vib to the server, connect to it via SSH and install:
esxcli software vib install -v /vmfs/volumes/datastore1/500.04.V0.53-0003.vib

We reboot the server, connect to it again via SSH and make sure that the package is installed:
esxcli software vib list | grep -i lsi

Now in the «Health Status» section, we can observe the status of the LSI MegaRAID controller:

Automation of sensor polling

But, of course, this is not enough to organize automatic monitoring because we won't know about the failure until we start the VMware vSphere client. It is necessary to automate the procedure for sensor polling. To do this, we use the “check_esxi_hardware.py” script, available at http://www.claudiokuenzler.com/nagios-plugins/check_esxi_hardware.php It is mainly an extension for Nagios, but it is quite universal and easily connected to other monitoring systems.

The script is written in the Python programming language and requires the PyWBEM library. Under Linux Debian and Ubuntu, it is installed through the standard system repositories:
apt-get install python-pywbem

The syntax for running «check_esxi_hardware.py» is quite simple:
check_esxi_hardware.py -H XXX.YYY.WWW.ZZZ -U root -P XXXXXXXX

In response, you will receive a brief report on the health of the server:
OK - Server: Supermicro X9DR3-F s/n: 0123456789 System BIOS: 3.0a 2013-07-31

You can make sure that the script polls the status of all sensors, including the LSI MegaRAID controller, by turning on detailed output of information:
check_esxi_hardware.py -H XXX.YYY.WWW.ZZZ -U root -P XXXXXXXX -v

The disadvantage of the script is that in order to poll the sensors, it needs the authorization data of the hypervisor administrator. It may not necessarily be root, but it must have the appropriate rights, otherwise it will not poll the sensors.

Let's try to simulate the failure of one of the HDDs. We reboot the server and go to the controller's WebBIOS. Choose one of the hard drives:

Let's go to its properties:

And disable:

We load the hypervisor and in the VMware vSphere client we see that there really is a failure:

And here is what "check_esxi_hardware.py" produces:

Did you like the article?