WilliamLam.com

Update on ESXi on Apple Mac Mini 2018 & Mac Pro 2019

by // 80 Comments

Although there has not been any news in some time regarding the support for ESXi on the latest Apple Mac Mini 2018 and the recently released Apple Mac Pro 2019, there has definitely been work happening behind the scenes at VMware. Today, I would like to share a pretty significant update as a result of some of these efforts.

MacOS Guest

One of the biggest issue which I had observed when using a T2-based Apple system with ESXi is that it would fail to boot a MacOS Guest and just keep rebooting the VM. I am very happy to announce that this issue has been resolved and ESXi can now properly recognize the Apple System Management Controller (SMC) device which is used as part of the MacOS Guest start up process. This now means a MacOS Guest will be able to properly boot on a T2-based Apple system.

Thunderbolt 3

Another impact of a T2-based Apple system with ESXi is that storage and networking devices connected to the Thunderbolt 3 ports are not visible. I am also happy to announce that this issue has been resolved and ESXi can now see PCIe devices that are attached to the Thunderbolt 3 ports.

An ESXi Advanced Setting change is required for Thunderbolt 3 to work correctly and the following command will need to be executed after installing ESXi:

esxcli system settings kernel set -s pciExperimentalFlags -v 16

Once the setting has been applied, a system reboot will be required and your PCIe devices will show up properly. In future, this additional configuration may not be required and can be detected based on the underlying hardware.

Both of the fixes mentioned above are included in the latest ESXi 6.7 Patch 02 (ESXi670-202004002) release which is available today! Hopefully this was the news that many of you have been waiting for 😀

UPDATE (09/02/21) - Per this official blog post, VMware will no longer pursue hardware certification for the Apple 2019 Mac Pro 7,1 for ESXi.

UPDATE (02/23/21) - The Community NVMe Driver for ESXi Fling now enables access to the local Apple NVMe device.

UPDATE (08/27/20) - The Apple 2018 Mac Mini 8,1 is now officially supported with ESXi 6.7 Update 3 which requires the latest ESXi 6.7 Patch 03 which also incorporates automatically setting the ESXi Advanced Setting for Thunderbolt 3 access.

UPDATE (06/25/20) - The Apple 2018 Mac Mini 8,1 is now officially on the VMware HCL and is fully supported with ESXi 7.0b, which contains the fixes mentioned above. See note below on 06/23 for more information.

UPDATE (06/23/20) - ESXi 7.0b has just been released and contains fixes for both the MacOS guest boot issue support for Thunderbolt 3 devices which now enables support for the vSphere 7 release. One additional enhancement, customers no longer need to configure the ESXi Advanced Setting to enable Thunderbolt 3 support, this is now automatically configured based on detecting an Apple hardware system such as an Apple Mac Mini 2018 or Apple Mac Pro 2019. This is a patch release and you will need to go to the VMware Patch Portal site to download and apply the update.

Now, before you rush out to start deploying MacOS Guests on either the Mac Mini or Mac Pro, I do have to mention that neither the Mac Mini 2018 or the Mac Pro 2019 will be officially supported by VMware. Due to the current situation that we are all in with COVID-19, personnel access to VMware facilities like many other organizations has been severely restricted and/or prohibited. In fact, much of the early validation was done by yours truly using a Mac Mini 2018 which I had access to (Thanks Michael Roy) as Engineering did not have access to hardware during the shelter in place orders. This also means that certifications of these platforms is still on-going and until these systems are officially listed on VMware's HCL, they will not be officially supported by VMware.

Disclaimer: VMware currently does not officially support the Apple 2019 Mac Pro7,1

[Read more...]

How to deploy the vCenter Server Appliance (VCSA) with a custom MAC Address?

by // 8 Comments

I recently had a question that came in from our field where a customer needed to deploy the vCenter Server Appliance (VCSA) with a specific MAC Address which was a requirement to ensure property connectivity within their network. This type of network requirement is not really new or unique, it is a common practice used to ensure only valid VMs with a static DHCP reservation can actually connect to a specific network but it certainly was the first time I had heard of this request for the VCSA.

With the default VCSA installer workflow, there is currently not a way to modify the network MAC Address which is automatically generated after the deployment of the OVA. Having said that, I have spent quite a bit of time exploring the various non-standard methods of deploying the VCSA in the past (see here, here and here) and with that information, you definitely can affect the MAC Address while still maintaining a valid VCSA deployment. With a bit of trial/error, there are two options depending if you are deploying the VCSA directly to an ESXi host (for initial setup) or to an existing vCenter Server. To demonstrate how this works, I have created a basic shell script called VCSAStaticMACAddress.sh which you can easily adapt to for a Windows-based environment.

The trick is that when you deploy to a vCenter Server endpoint, the required OVF properties are persisted which would allow you to only deploy the VCSA but not actually power it on and there you can easily augment a number of settings including the MAC Address. In the case of deploying directly to an ESXi host, OVF properties are not persisted and hence a challenge if you wish to make changes prior to powering on the VM. In earlier versions, it was possible to set these OVF properties by way of using the extraConfig property of the VM but it looks like this trick no longer works and requires a slight variation of the workflow which is described in the instructions below.

[Read more...]

How to automate the creation multiple routable VLANs on single L2 network using VyOS

by // 5 Comments

My personal homelab has a very simple network topology, everything is connected to a single flat network. This has served me well over the years, but sometimes it can prevent me from deploying more complex scenarios. Most recently while working with NSX-T and Project Pacific, I had a need for additional VLANs which my home router does not support. There are a number of software solutions that can be used including the popular pfSense, which I have used before.

Over the Winter break, a colleague introduced me to VyOS, which is another popular software firewall and router solution. I had not heard of VyOS before but later realized it was derived from Vyatta, which I had heard of, but development of that solution had stopped and VyOS is now the open source version of that software. Having never played with VyoS before, I thought this might be a good learning opopournity and started to dabble with VyOS over the holiday. At a high level, I have VyOS connected to two networks: Outside network as VyOS refers which is your local LAN and Inside network as VyOS refers which is an is an isolated vSphere Portgroup (VSS/VDS) that is not connected to anything and configured to pass all traffic (4095). From here, you can create multiple VLANs in VyOS which can then be untagged using Virtual Guest Tagging (VGT) by placing a Nested ESXi VM on the same isolated portgroup and then creating the respective portgroups within the Nested ESXi VM mapping to the VyOS VLANs you have created.

One of the nice benefits of this solution is that you can create multiple "Isolated" yet routable networks that can still reach your primary LAN network and still have  to access core infrastructure services running like Active Directory, DNS, etc. which was one of my requirements.  After figuring out how VyOS works and applying that to my specific use case, I thought why not build some basic automation to setup this solution as I probably will forget how I setup everything. Initially I was using the VyOS OVA but later found out it was an extremely out of date there was no public version of the latest VyOS release in OVA form. I decided to use their latest rolling release and apply some vSphere API Automation to not only install VyOS but also fully configure based on template containing VyOS commands. I know the latest version of VyOS now includes a REST API but its a bit of a chicken/egg to enable and not very friendly to use compared to the solution I have built.

[Read more...]