Starting with Version 2.2 of Zabbix, we can now monitor VMware hosts. From What’s new in Zabbix 2.2.0:

A new feature in Zabbix 2.2.0 is VMware virtual machine monitoring. It allows to monitor VMware vCenter and vSphere installations for various VMware hypervisor and virtual machine properties and statistics.

Zabbix can use low-level discovery rules to automatically discover VMware hypervisors and virtual machines and create hosts to monitor them, based on pre-defined host prototypes. See Virtual machine monitoring for more detailed information.

So I decided to update Zabbix to 2.2 and Monitor VMware hosts.

Backup current Zabbix Configuration

All the settings for zabbix are stored under /etc/zabbix, so let’s backup that directory:

sudo tar cpvjf /tmp/zabbix-conf-back.tar.bz2 /etc/zabbix

If you have the space, backup the MySQL DB as well:

mysqldump -h localhost -u zabbix -p zabbix > zabbix.sql

Update Zabbix to 2.2

I was using Debian so the process is pretty easy. First stop the service:

sudo service zabbix-server stop
sudo service zabbix-agent stop

Next get the new repository:


Then install the new deb package and update all the packages:

sudo dpkg -i zabbix-release_2.2-1+wheezy_all.deb
sudo apt-get update
sudo apt-get dist-upgrade

During the install process it will ask you to setup a Database, skip that. The install process will also install new configuration files and will ask you which files you want to keep, choose the following:

keep the maintainer version

There were a lot of new options and I decided to just keep the new configuration and re-add my original settings into the new version. After the install, the zabbix services are automatically started, so let’s go ahead and stop them:

sudo service zabbix-server stop
sudo service zabbix-agent stop

Then fix the configuration to include your original settings, here is what I had after I was done:

$sudo grep -Ev "^#|^$" /etc/zabbix/zabbix_server.conf

Then do the same thing for the agentd configuration, here is what I had after I was done:

$sudo grep -Ev "^#|^$" /etc/zabbix/zabbix_agentd.conf

The new version installs it’s own configuration into apache automatically:

$ls -l /etc/apache2/conf.d/zabbix
lrwxrwxrwx 1 root root 23 Feb  8 12:41 /etc/apache2/conf.d/zabbix -> /etc/zabbix/apache.conf

If you had any apache settings for the frontend you can remove them and just use that configuration. Restart apache after you removed (if any) the settings for the zabbix frontend:

sudo service apache2 restart

Then let’s copy our original PHP frontend settings into the new install:

sudo cp /etc/zabbix/zabbix.conf.php /etc/zabbix/web/.

Then start the zabbix services:

sudo service zabbix-server start
sudo service zabbix-agent start

Check out the logs to make sure everything is okay:

less /var/log/zabbix/zabbix-server.log
less /var/log/zabbix/zabbix-agent.log

Then visit your PHP frontend (http://localhost/zabbix) and make sure all the information is still there.

Import VMware Templates

The new VMware Templates are available here. Go ahead and download them:


Upon initial import I received the following error:

Cannot find value map “VMware VirtualMachinePowerState” used for item “Power state” on “Template Virt VMware Guest”.

Looks like another person ran into a similar issue here (it’s in German, but you can always translate it) and the fix is to add a couple of values to the MySQL Database as described in this forum. Here are the commands that need to be run on the MySQL DB. When I ran the commands they failed, because I was using those values for something else, so I just incremented my valuemapid and it worked. Here are the commands I ended up running:

INSERT INTO `valuemaps` (`valuemapid`,`name`) values ('13','VMware status');
INSERT INTO `valuemaps` (`valuemapid`,`name`) values ('14','VMware VirtualMachinePowerState');
INSERT INTO `mappings` (`mappingid`,`valuemapid`,`value`,`newvalue`) values ('75','13','0','gray');
INSERT INTO `mappings` (`mappingid`,`valuemapid`,`value`,`newvalue`) values ('76','13','1','green');
INSERT INTO `mappings` (`mappingid`,`valuemapid`,`value`,`newvalue`) values ('77','13','2','yellow');
INSERT INTO `mappings` (`mappingid`,`valuemapid`,`value`,`newvalue`) values ('78','13','3','red');
INSERT INTO `mappings` (`mappingid`,`valuemapid`,`value`,`newvalue`) values ('82','14','0','poweredOff');
INSERT INTO `mappings` (`mappingid`,`valuemapid`,`value`,`newvalue`) values ('83','14','1','poweredOn');
INSERT INTO `mappings` (`mappingid`,`valuemapid`,`value`,`newvalue`) values ('84','14','2','suspended');

After that, I imported the templates in this order:

  1. Template_Virt_VMware_Hypervisor-2.2.0.xml
  2. Template_Virt_VMware_Guest-2.2.0.xml
  3. Template_Virt_VMware-2.2.0.xml

Enable the VMware Collector on the Zabbix Server

Now that we have the templates in place, let’s enable the VMware collector. First make sure the server is compiled with the libxml2 option. From Virtual machine monitoring:

For virtual machine monitoring to work, Zabbix should be compiled with the –with-libxml2 and –with-libcurl compilation options.

This can be confirmed, by checking the linked libraries of the binary. Here is what I saw:

elatov@kerch:~$ldd /usr/sbin/zabbix_server | grep -E 'xml|curl' => /usr/lib/x86_64-linux-gnu/ (0x00007fd56965a000) => /usr/lib/x86_64-linux-gnu/ (0x00007fd567f29000)

Then to enable the collector I added the following to my /etc/zabbix/zabbix_server.conf file:

StartVMwareCollectors = 1
VMwareCacheSize = 8M
VMwareFrequency = 60

I then restarted the service:

sudo service zabbix_server restart

and I saw the following in the logs under /var/log/zabbix/zabbix_server.log:

17807:20140418:135649.762 Starting Zabbix Server. Zabbix 2.2.3 (revision 44105)
 17807:20140418:135649.762 ****** Enabled features ******
 17807:20140418:135649.762 SNMP monitoring:           YES
 17807:20140418:135649.762 IPMI monitoring:           YES
 17807:20140418:135649.762 WEB monitoring:            YES
 17807:20140418:135649.763 VMware monitoring:         YES
 17807:20140418:135649.763 Jabber notifications:      YES
 17807:20140418:135649.763 Ez Texting notifications:  YES
 17807:20140418:135649.763 ODBC:                      YES
 17807:20140418:135649.763 SSH2 support:              YES
 17807:20140418:135649.763 IPv6 support:              YES
 17807:20140418:135649.763 ******************************
 17842:20140418:135649.798 server #28 started [vmware collector #1]

Enable VMware Discovery

To Discover the Hypervisor and VMs, create a new host and leave the default agent interface:


Then under Macros, define the following:

  • {$URL} - VMware service (vCenter or ESX hypervisor) SDK URL (https://servername/sdk).
  • {$USERNAME} - VMware service user name
  • {$PASSWORD} - VMware service {$USERNAME} user password


Lastly assign the VMware Template to it:


After some time, you will see the VMs and Hypervisors discovered and their corresponding templates assigned to them. For example here are two VMs that I was running on one of my hosts:


If you go inside the VM, you will see that most of the items are grayed out:


and the host is the UUID of the VM. Same thing was done for the hypervisor, except it attached the VMware Hypervisor Template to it, instead of the VMware Guest one.

Expanding the VMware Templates

By default the templates didn’t have any graphs defined so I ended up defining graphs for CPU Usage and Memory Usage. I just ended up plotting the already existing items and that was pretty easy. I also ended up creating calculated items to help out with some triggers. Here are the calculated items, I added to the Hypervisor Template:


Here is what I had in the HV CPU Usage Percentage:


Here is actual formula:


I decided to use vmware.hv.hw.cpu.threads to calculate total CPU, but vmware.hv.hw.cpu.num is also available (I thought it was appropriate since HyperThreading is enabled).

All of the available VMware keys, are listed here. For creating calculated items, check out the zabbix documentation Calculated items. And here was the triggers I ended up creating:


All the trigger functions are defined in Supported trigger functions. I also ended up modifying the discovery rules to do graphing for the Datastores and to trigger on high latency. I ended up making similar modifications to the VMware Guest Template. BTW the default keys for the VMware Guest are the following:


It’s not much but it’s a good start. There are already a couple of feature request to get more data. For example here is a link to one asking for the %RDY of the VM. Now I can tell my ESX host is completely under utilized :)