Yes you can. Have a look at the ESX/ESXi Discovery CET (Community Enablement Tutorial) for a bit more information on the ESX discovery process, how to setup credentials and inportant documentaion to read before you get started.
The document is very interesting but it is not so clear if through the VCenter you can also discover the details (like processes, network connections and other informations) related to the Virtual Machines hosted in the ESX/ESXi .
Someone knows if in the ADDM 8.3.2 you must discover the Virtual Machine directly via ssh or you can also discover them through the Vcenter?
I do mention this, but yes I see that it isn't very clear so hopefully this answer will help.
vCenter/vSphere are credentials used to interrogate the ESX host itself (so the host container).
When ADDM discovers an ESX host it records all sorts of details about the ESX host itself and the virtual hosts it manages but from a logical perspective (see slide 20).
If you want to see the discovered processes, network connections and other information that are found running on the VM, that happens when the VM is discovered directly. If the VM is Linux, ssh is typically used. If the VM is Windows, then Windows discovery methods are used (see slide 4, first bullet point).
So to directly answer your question, yes you must discover the VM directly in order to see specific details about that VM.
I know you didn’t ask this question, but I’ll provide it just in case it isn’t obvious. The serial number of the VM is used to make the host container/contained host connection. So if there is a problem collecting the serial number on the VM…then the relationship won’t be made. This connection can be seen on the host node page of the ESX host (see image below).
I have a question related to VCenter discovery partly related to the above. The serial number of the VMs can change under certain circumstances even thought the 'server' itself retains it's attributes/role in the infrastructure. How does ADDM reconcile this when it happens and avoid creating duplicates in its own repository (assuming it is using serial number as part of its identification strategy - perhaps that's an invalid assumption), and in particular then how do we avoid creating duplicates in Atrium CMDB when we sync??
Sure I’m happy to answer another question, however I’m not sure it is related to vCenter (but you can correct me if I’m wrong). I only bring this up so I don’t confuse other folks who may read this by searching for vCenter.
Yes you are correct, serial number is one of the attributes used in the host identification algorithm but so are things like IP address, and the OS of the host. You can go here to read more on how ADDM uniquely indentifies a host.
What are the certain circumstances that you mention? I’m not a VMWare admin, but when I make a copy of one of my VMs and then start it up, I get a dialog box that asks me if I moved the VM or if it is a copy. If I select “moved” the serial number remains the same, if I select “copy” the serial number changes. Is this what you mean? Or is there something specific to VMware vCenter administration that I’m not aware of?
BTW I'm just following up that I have updated the wording to the ESX/ESXi Discovery CET that will hopefully better explain what is needed for both ESX host and VM host discovery.
Thanks very much Giorgio for the feedback and helping me make the Atrium Discovery CET's better.
Thanks for the reply. You're right, this is not specifically about VCenter, it's more about understanding the strategy ADDM uses for identifying virtual servers and your link helped me understand that. We have several discovery sources as well as ADDM and where there is a chance more than one may be contributing the same device, it helps to understand what the identification approach is to avoid nasty surprises at recon. time.
The link for ESX & ESXi discovery through ADDM is excellent.
Zoe and Carey, there is one more question which is asked to us by the clients often. If there are 3 VMs setup on 1 ESX -> VMWare machine, then how does ADDM identify the 3 separate VMs..? Or does it Identify all 3 as a single VM machine..? How many separate CIs will be available in CMDB after the ADDM -> CMDB sync is performed..?
Please provide your thoughts.
You are correct. ADDM would create 4 "host nodes" (we use different terminology ):
1 for the ESX host (the host container)
and 3 for the VM's
This is assuming of course you had good credentials to scan all 4 devices becasue they would all need to be discovered directly. And yes they would sync to the CMDB as 4 separate CI's.