VMware Hypervisor Template - No Graphs

[rpi_dwillis77]

Hello,

I have setup VMware monitoring using the built-in VMware FQDN template (added vCenter via FQDN and setup the necessary permissions for Zabbix which then discovered hosts and VMs). Mostly everything is good, however for some reason with the hosts I don't have any graphs (when I view an ESXi host that was discovered through vCenter on the Monitoring -> Hosts page, it has data under Latest Data but no graphs - Graphs is greyed-out). If I manually look at the data then I can view a graph for each individual one but it doesn't have the big list of graphs like we have for most hosts which have data. Is there a reason for this - something I missed in the setup maybe? Or is that just how this template behaves? And if that is the case - are there any workarounds?

Any help is much appreciated, thanks!

[rpi_dwillis77]

Hello,

After digging around and researching this myself I came to the conclusion that the VMware Hypervisor template is just somewhat limited when it comes to what data it displays. I cannot explain why the "graphs" link is disabled in the Monitoring -> Hosts page though, I would think it would still be enabled to at least show the few graphs that are available (for the pieces of information that the VMware Hypervisor template does collect about the host). The only way I've found to view such data is as noted in my post above (to manually go to the host's Latest Data and view the graph for each piece of data individually). I will say however that the VM Guest templates that it uses for guest VMs are great and provide lots of good info.

So having accepted the fact that the hypervisor template is just subpar when it comes to data gathering, I started looking for other ways to get more detailed information about the hosts, and after trying a few different ways including IPMI/iDRAC, SNMP, etc., the easiest method I found was SNMP - this provides a wealth of great information and a ton of graphs to go with it.

Note the information below assumes you've already setup the basic VMware monitoring via the VMware FQDN template and let the discovery run to discover your VMs and hosts. This process is described here, here and here, among other places.

To setup the SNMP monitoring, I first configured the hosts for SNMP - this must be done via CLI, either esxcli directly on the host if you have SSH or console access or it can be done remotely through PowerCLI. Just enable it, set your community string and optionally your location and contact info. Next I modified the host prototype associated with the "discover VMware hypervisor" discovery rule to include a SNMP interface (this is necessary because although we can link additional templates to discovered hosts manually, we cannot add additional interfaces to discovered hosts so we must do this by modifying the prototype). This is better in the long run anyway, if you want to use this method to monitor all your hosts because it means all discovered hosts will get the same configuration. So I went into Configuration -> Templates -> VMware FQDN -> Discovery Rules tab -> Discover VMware hypervisors -> Host Prototypes tab, then click on the {#HV.NAME} prototype there to modify it. In here, add a new interface of type "SNMP", choose to monitor by IP on port 161 (assuming you aren't using a nonstandard port on the ESX side) and the value should be the macro {#HV.IP} (which will expand to your host's IP). The other change you need to make here is to link the Linux by SNMP template (in addition to the default VMware Hypervisor template that is linked to the prototype by default). Once done, hit Update to save the changes.

Now to get the host prototype changes to take effect on existing hosts (since it doesn't appear that changes made to a prototype will push down to hosts that were already previously discovered and created from that prototype), I went to the definition of the Discover VMware Hypervisor discovery rule and unchecked the "discovery" box there (make sure your "keep lost resources period" is not 0d so you don't lose your existing data for your hosts) and then waited for that discovery to run again to "un-discover" the existing hosts (or you can force-execute it if you want to speed up the process). You'll know the hosts have been "un-discovered" once you see little orange boxes next to your ESXi hosts in the host list that says they will be "removed in X number of days". Once this had occurred I went back into the Discover VMware hypervisor discovery rule, re-enabled the "discovery" box, saved, and let the discovery run again to re-discovery the hosts - and this will push the changes you made to the prototype down to them as well. Now every host should have an SNMP interface attached to it as well as the Linux by SNMP template - now just wait a few hours for the data collection to begin and you should start to see all types of data about your hosts including detailed per-interface metrics for all network interfaces (which is really what I was after), etc.

Hope this may help anyone in a similar situation!

Thanks!

[rpi_dwillis77]

Hello,

After digging around and researching this myself I came to the conclusion that the VMware Hypervisor template is just somewhat limited when it comes to what data it displays. I cannot explain why the "graphs" link is disabled in the Monitoring -> Hosts page though, I would think it would still be enabled to at least show the few graphs that are available (for the pieces of information that the VMware Hypervisor template does collect about the host). The only way I've found to view such data is as noted in my post above (to manually go to the host's Latest Data and view the graph for each piece of data individually). I will say however that the VM Guest templates that it uses for guest VMs are great and provide lots of good info.

So having accepted the fact that the hypervisor template is just subpar when it comes to data gathering, I started looking for other ways to get more detailed information about the hosts, and after trying a few different ways including IPMI/iDRAC, SNMP, etc., the easiest method I found was SNMP - this provides a wealth of great information and a ton of graphs to go with it.

Note the information below assumes you've already setup the basic VMware monitoring via the VMware FQDN template and let the discovery run to discover your VMs and hosts. This process is described here, here and here, among other places.

To setup the SNMP monitoring, I first configured the hosts for SNMP - this must be done via CLI, either esxcli directly on the host if you have SSH or console access or it can be done remotely through PowerCLI. Just enable it, set your community string and optionally your location and contact info. Next I modified the host prototype associated with the "discover VMware hypervisor" discovery rule to include a SNMP interface (this is necessary because although we can link additional templates to discovered hosts manually, we cannot add additional interfaces to discovered hosts so we must do this by modifying the prototype). This is better in the long run anyway, if you want to use this method to monitor all your hosts because it means all discovered hosts will get the same configuration. So I went into Configuration -> Templates -> VMware FQDN -> Discovery Rules tab -> Discover VMware hypervisors -> Host Prototypes tab, then click on the {#HV.NAME} prototype there to modify it. In here, add a new interface of type "SNMP", choose to monitor by IP on port 161 (assuming you aren't using a nonstandard port on the ESX side) and the value should be the macro {#HV.IP} (which will expand to your host's IP). The other change you need to make here is to link the Linux by SNMP template (in addition to the default VMware Hypervisor template that is linked to the prototype by default). Once done, hit Update to save the changes.

Now to get the host prototype changes to take effect on existing hosts (since it doesn't appear that changes made to a prototype will push down to hosts that were already previously discovered and created from that prototype), I went to the definition of the Discover VMware Hypervisor discovery rule and unchecked the "discovery" box there (make sure your "keep lost resources period" is not 0d so you don't lose your existing data for your hosts) and then waited for that discovery to run again to "un-discover" the existing hosts (or you can force-execute it if you want to speed up the process). You'll know the hosts have been "un-discovered" once you see little orange boxes next to your ESXi hosts in the host list that says they will be "removed in X number of days". Once this had occurred I went back into the Discover VMware hypervisor discovery rule, re-enabled the "discovery" box, saved, and let the discovery run again to re-discovery the hosts - and this will push the changes you made to the prototype down to them as well. Now every host should have an SNMP interface attached to it as well as the Linux by SNMP template - now just wait a few hours for the data collection to begin and you should start to see all types of data about your hosts including detailed per-interface metrics for all network interfaces (which is really what I was after), etc.

Hope this may help anyone in a similar situation!

Thanks!