VMware Fusion and Workstation continues to be another popular way for customers to run a VMware Homelab while leveraging a users existing desktop. In the early days of vSphere 6.5, the method to deploy the vCenter Server Appliance (VCSA) to Fusion/Workstation was less than ideal with a lot of manual steps. In 2017, the Fusion/Workstation team introduced native OVF support and that made deploying the VCSA much simpler, especially with the VCSA two stage installer.
Even though this is not an officially supported method from VMware for deploying the VCSA, the process has not changed for the last several releases and it just works which was great for our users. With vSphere 8, it looks like there has been a change to the VCSA installer that causes a failure during the Stage 2 configuration.
I was made aware of this issue from a colleague who was looking to setup a vSphere 8 environment using VMware Workstation.
Taking a look at the log file, he noticed the error was regarding the upgrade.import.directory OVF property not being set, which seems to be causing the issue.
# cat /var/log/firstboot/vlcm_firstboot.py_9117_stderr.log
Traceback (most recent call last):
File "/usr/lib/vmware-vlcm/firstboot/vlcm_firstboot.py", line 167, in <module>
main()
File "/usr/lib/vmware-vlcm/firstboot/vlcm_firstboot.py", line 129, in main
fb = VlcmFirstboot(VLCM_COMP, VLCM_SOLUSERNAME)
File "/usr/lib/vmware-vlcm/firstboot/vlcm_firstboot.py", line 50, in __init__
os.path.join(get_install_parameter('upgrade.import.directory'), "vlcm")
File "/usr/lib/vmware/site-packages/cis/tools.py", line 88, in get_install_parameter
raise InstallParameterException('Install-parameter %s not set' % param)
cis.exceptions.InstallParameterException: {
"detail": [
{
"id": "install.ciscommon.internal.error",
"translatable": "Encountered an internal error.\n\n%(0)s",
"args": [
"Install-parameter upgrade.import.directory not set"
],
"localized": "Encountered an internal error.\n\nInstall-parameter upgrade.import.directory not set"
}
],
"componentKey": null,
"problemId": null,
"resolution": {
"id": "install.default.failure.resolution",
"translatable": "This is an unrecoverable error, please retry install. If you encounter this error again, please search for these symptoms in the VMware Knowledge Base for any known issues and possible resolutions. If none can be found, collect a support bundle and open a support request.",
"localized": "This is an unrecoverable error, please retry install. If you encounter this error again, please search for these symptoms in the VMware Knowledge Base for any known issues and possible resolutions. If none can be found, collect a support bundle and open a support request."
}
}
Using one of my previous blog post for guidance, he discovered a quick workaround to the problem by simply ensuring this variable is configured with a default value. After running into the problem myself and verifying the solution, I figure this might be useful for anyone looking to run vSphere 8 using VMware Fusion or Workstation, so here are the instructions to work around this issue.
Disclaimer: This is not officially supported by VMware, use at your own risk
Step 1 - Extract the contents of the VCSA 8.0 ISO to your local desktop. If you are on a macOS system, you may also need to remove security quarantine flag from the extracted directory and ensure the directory has write access (e.g. chmod 755).
Step 2 - Change into the vcsa directory, which you see the VCSA OVA. Run the following command to convert the VCSA OVA to an OVF
ovftool VMware-vCenter-Server-Appliance-8.0.0-XXX_OVF10.ova VMware-vCenter-Server-Appliance-8.0.0-XXX_OVF10.ovf
Note: If you do not have OVFTool installed, there is a local version bundled in the ovftool directory and you run that by specifying the local path to the OVFTool utility.
Step 3 - Delete the original VCSA OVF manifest file so that validation will not fail due to our changes.
rm VMware-vCenter-Server-Appliance-8.0.0-XXX_OVF10.mf
Step 4 - Edit the VCSA OVF file and locate the following section and change the value of ovf:userConfigurable to true and then save your changes.
<Property ovf:key="guestinfo.cis.upgrade.import.directory" ovf:type="string" ovf:userConfigurable="true" ovf:value="/storage/seat/cis-export-folder"> <Label>Upgrade Destination Export Folder</Label> <Description>Folder where exported source data will be stored in the appliance. Optional. Set only for upgrade</Description> </Property>
At this point, we are now ready to deploy our modified VCSA OVF using either VMware Fusion or Workstation. You can also delete the OVA as that is no longer needed and will give you back some space.
In my setup, I am using VMware Fusion and once the import of the VCSA OVF has completed, you will only need to fill out the networking and the password section for both SSO and the OS as shown in the screenshot below. If are going to use DHCP, then you only need to fill out two properties along with the passwords and then click continue.
The VCSA will now begin its bootstrap process for Stage 1 and this can take a few minutes depending on your available resources which will also include a VM reboot. Be patient and wait until the VCSA displays the DCUI gray/blue screen with IP Address that you had specified earlier.
Once you have the IP Address of the VCSA, open a browser to the following URL: https://[VCSA-IP]:5480 which is the VCSA VAMI interface for configuring Stage 2.
Click on the Setup icon to begin the VCSA Stage 2 configuration wizard and fill out the required fields as shown in the screenshots below.
Finally, click continue to start the Stage 2 configuration and after ~5-10 minutes, you will have a fully configured VCSA 8.0 which you can connect to vSphere UI using either the IP Address or hostname that you had specified earlier.
Hello to ALL:
We installed VCSA 6.7 as a VM on Vmware Workstation using the method VMware recommends. It installed without any issues and worked perfectly.
We then installed VCSA 7.0 the same way without issues. But when we tried to install VCSA 7 U3 it would fail during stage 2 of the install process. So we installed VCSA 7 U3 on a virtual ESXi 7 Server (nested). The install finished without any issues. Then we opened Vmware Workstation and conected to the ESXi 7 Hypervisor. We then saw the virtual machines that were installed using the ESXi 7 Hypervisor and we selected the VCSA 7 U3 Virtual Machine and there is a download option so we selected the download option and downloaded all the Virtual Machine files to a directory on our Windows 11 PC. Then we opened the .VMX file with Vmware Workstation and started this VCSA 7 U3 virtual machine and all worked perfectly.
Yes this is somewhat time consuming but since we could not install VCSA 7 U3 on Vmware Workstation without error using the Vmware's suggested method, we did this workaround and now we have VCSA 7 U3 installed as a virtual machine on our Vmware Workstation.
Great article. Will this be fixed in the next vCenter release update?
No, since deploying to Fusion/Workstation isn’t officially supported for the VCSA 🙁
We converted the .OVA to an .OVF and modified the .OVF as suggested and the Install of the VCSA on Vmware Workstation worked perfectly.
Thanks to Mike Vilosaw.
I follow the guide deploy vcsa8 in fusion12, but the vm network was always DHCP, and root password was not set. the ovf configurations dose not passed into vm. Are there some additional configuration？
The install went properly for me in VMware workstation. The issue I have is when I tried to launch the vsphere 8 client, it states that it cannot reach page. I've checked everything: DNS records, disabled firewall and antivirus, cleared browser cache, etc.... but nothing works. What is weird is that the vsphere 7 client runs fine.
I have same issue to deploy some hardware.
I setup vcenter8 manay time on esxi8.
it got error on 68%.
how can I fix it?
Traceback (most recent call last):
File "/usr/lib/vmware-vlcm/firstboot/vlcm_firstboot.py", line 167, in
main()
File "/usr/lib/vmware-vlcm/firstboot/vlcm_firstboot.py", line 129, in main
fb = VlcmFirstboot(VLCM_COMP, VLCM_SOLUSERNAME)
File "/usr/lib/vmware-vlcm/firstboot/vlcm_firstboot.py", line 50, in __init__
os.path.join(get_install_parameter('upgrade.import.directory'), "vlcm")
File "/usr/lib/vmware/site-packages/cis/tools.py", line 88, in get_install_parameter
raise InstallParameterException('Install-parameter %s not set' % param)
cis.exceptions.InstallParameterException: {
"detail": [
{
"id": "install.ciscommon.internal.error",
"translatable": "Encountered an internal error.\n\n%(0)s",
"args": [
"Install-parameter upgrade.import.directory not set"
],
"localized": "Encountered an internal error.\n\nInstall-parameter upgrade.import.directory not set"
}
],
"componentKey": null,
"problemId": null,
"resolution": {
"id": "install.default.failure.resolution",
"translatable": "This is an unrecoverable error, please retry install. If you encounter this error again, please search for these symptoms in the VMware Knowledge Base for any known issues and possible resolutions. If none can be found, collect a support bundle and open a support request.",
"localized": "This is an unrecoverable error, please retry install. If you encounter this error again, please search for these symptoms in the VMware Knowledge Base for any known issues and possible resolutions. If none can be found, collect a support bundle and open a support request."
}
}
I follow the article to make change. It cannot complete to setup
Is this actually a Fusion/Workstation question or ESXi? If the latter, you do NOT need to deploy OVA manually but rather use the actual VCSA Installer
i tried to deploy on ESXi8 on R710.
I could install vcenter 8.0 beta. But I could install GA.
I just found out. It is DNS issue.
I need to put IPv4 PTR in DNS server.
What exactly have you done ?
I've done the DNS configuration but it still does not work
Hi William,
Thank you for the write-up! I am experiencing the same error as you described above, but while trying to install on ESXi instead of Workstation/Fusion. Could you advise me on how to deploy the modified OVA through the VCSA installer on ESXi? I attempted to do so after making the modifications mentioned above, but I received an error message stating that the installer was unable to validate the OVA file. I attempted this yesterday, but forgot to note down the exact error message. I am currently trying to reproduce the error.
If the above method does not work, is it possible to log in to the appliance via SSH after the first stage has completed but before the second stage has started? This way, I could make modifications then instead of modifying the OVA. I have been troubleshooting this issue for about a week now and I am unsure how to move forward. Does anyone know why this error is thrown in the first place?
If you have ESXi, WHY are you using this method of deployment? The VCSA provides native UI/CLI installer that can deploy to ESXi or vCenter Server endpoint which will give you best experience. The workaround mentioned here is ONLY applicable if you're not using ESXi/vCenter and want deploy to Fusion/Workstation which doesn't have support for VCSA Installer. Please use the VCSA Installer
Thanks for the quick response. To clarify, I am receiving the same error you mentioned above while attempting to install via the VCSA installer on my ESXi host during the second stage. That is why I was attempting to modify the OVA as per your suggestion.
I was able to replicate the issue again and here are a couple of screenshots of what I am referring to: https://imgur.com/a/ftJ9U7g. It looks like both Stephen Yu and myself have experienced the same issue while trying to deploy the VCSA to an ESXi host via the GUI installer.
Best to file an SR, I've not ran into this issue nor heard of this but could be specific version of ESXi or something else is going on
I figured out the solution! After reading Stephen Yu's comment about a DNS issue and the need to add an IPv4 PTR to the DNS server, I redeployed the VCSA using the GUI installer. Before doing so, I made sure to add DNS entries to the hosts file on my Windows VM. I added entries for the ESXi host I was installing the VCSA on and for the VCSA itself. During the VCSA deployment, I only included my DNS server and did not include 1.1.1.1. After completing stage 1, instead of navigating to the VCSA on port 5480 in the browser to continue installation, I attempted to complete stage 2 via the GUI installer on my Windows VM. These changes resulted in a successful installation. I'm not sure which change fixed the issue or if it was a combination of them all, but I hope this helps someone else in the future.
Thanks. Helped my deploy one over the weekend.
Thanks for the article. Worked like charm!!
You will need to add the following line to the VMX file if using the guestinfo options William documented in https://williamlam.com/2016/10/how-to-deploy-the-vcenter-server-appliance-vcsa-6-5-running-on-vmware-fusion-workstation.html.
guestinfo.cis.upgrade.import.directory = ""
It worked! you are number 1! 🙂
Is VSphere 8 and VCenter 8 already a ripe product ? I don't think so!
I'm working since many years with vmware products. Never had such issues as with version 8!
Using VCSA 6.7 U3 for several years now. The installation was easy and it is working fine. Before was using the vcenter on a windows-vm.
For a new cluster, was trying now the version 8 - try before you buy.
Imported the vcsa 8 OVA into the esxi 8 and got the error mentioned above.
Then tried several times with the modified OVF and it gets the same error. It stops on the "Starting VMwareupdate manager" and then stops.
It's a stupid installation, it should continue the installation of stage 2 as the update manager is not crucial in the beginning.
Interesting is, that even the stage 2 setup does not complete. The VCenter runs. But only until next boot. 🙁
I wont to avoid using again a windows vm with vcenter and database, but if the vcsa will not work that's the only way to go.
Any more ideas on that ? Or do we have to wait until version 8.1 is releaseed ?
I'm having the same issue, even after changing the ovf file.
Found the solution:
Instead of using the VCSA with the predefined vm. I've used the windows installer to let the installer create a vcenter vm on the ESXi server.
The installer run through the stage 1 and 2 without issues !
The login to the vcenter administration is working.
But, the login to the vcenter server was not able as it tries to connect to the vcenter domain server.
It may not be your windows domain and it requires a local DNS entry for the vcenter domain. i.e. I've used vsphere.local and created a DNS entry on my local dns server for vsphere.local with the ip-address of the vcenter virtual server.
This seems to be a new thing in version 7 or 8.
In my old vcenter 6.7 u3 this was not required.
https://changineer.info/vmware/hypervisor/vmware_vcsa_gui02.html
I have the same issue and i fixed it with your blog, thank you!
Hi William, Can you please help me. I'm using a macbook and trying to install VCSA 8.0 on fusion. I'm stuck at the installation of the stage 2 exactly as shown in your article. But, I'm not familiar with the mac command lines to exactly run the ovf tool. Please help me.
Well, you need to actually download and install the ovftool utility (https://developer.vmware.com/web/tool/4.5.0/ovf-tool) and then you can follow the remainder steps
Just a side comment, as I had also issues, listed above with this version of vcenter. Unbelievable - version 7 of a real stable product and then version 8 has so many issues. The quality of the application is not related to the real high price. I0ve bought the small package for 3 servers with vcenter. But you may not do a lot with this limited package.
It's time to move to Proxxmox, where you get the whole package including backup software for a very affordable support price. For me it's worth testing it
For me, the ovf-conversion did not really help; I still ran into the same issue.
However, mounting the iso and running the Win32-installer directly from the iso worked like a charm.
Everything worked except it still fails on Workstation 17 PRO with downloaded stderr page below. Tried 3 times same with all the proper changes. Any ideas?
=============
Traceback (most recent call last):
File "/usr/lib/vmware-vlcm/firstboot/vlcm_firstboot.py", line 167, in
main()
File "/usr/lib/vmware-vlcm/firstboot/vlcm_firstboot.py", line 129, in main
fb = VlcmFirstboot(VLCM_COMP, VLCM_SOLUSERNAME)
File "/usr/lib/vmware-vlcm/firstboot/vlcm_firstboot.py", line 50, in __init__
os.path.join(get_install_parameter('upgrade.import.directory'), "vlcm")
File "/usr/lib/vmware/site-packages/cis/tools.py", line 88, in get_install_parameter
raise InstallParameterException('Install-parameter %s not set' % param)
cis.exceptions.InstallParameterException: {
"detail": [
{
"id": "install.ciscommon.internal.error",
"translatable": "Encountered an internal error.\n\n%(0)s",
"args": [
"Install-parameter upgrade.import.directory not set"
],
"localized": "Encountered an internal error.\n\nInstall-parameter upgrade.import.directory not set"
}
],
"componentKey": null,
"problemId": null,
"resolution": {
"id": "install.default.failure.resolution",
"translatable": "This is an unrecoverable error, please retry install. If you encounter this error again, please search for these symptoms in the VMware Knowledge Base for any known issues and possible resolutions. If none can be found, collect a support bundle and open a support request.",
"localized": "This is an unrecoverable error, please retry install. If you encounter this error again, please search for these symptoms in the VMware Knowledge Base for any known issues and possible resolutions. If none can be found, collect a support bundle and open a support request."
}
}
=====
Never mind, found the problem and solved it.
Thank you very much. Just was able to install a vcenter 8.0 on a VMware Workstation PRO 17 installation in my home lab, which failed before with the mentioned "encountered an internal error. see /var/log/firstboot/vlcm_firstboot.py_9653_stderr.log" .
Hola y muchas gracias por tu aporte, si les ayuda yo encontre una solucion menos complicada, aca se los detallo. desde mi esxi seleccione crear maquina virtual y elegi la opcion implementar una maquina apartir de OVF u VMDK, le puse un nombre y agregue el archivo ova de la ruta vcsa de mi instalador. siguiente a todo, esperar unos minutos y cuando cargue el sistema hasta la pantalla azul con negro de vcsa8, debes ingresar para poner tu nueva clave, la clave por defecto es vmware.
luego ingresar al ip:5480 y pones la clave y comenzara a cargar la segunda etapa. exitos.
Followed the steps you outline, but am getting "Cannot find a peer process to connect to" when I try and run the VM for the first time. Any ideas?
Nevermind that error, it was the Security reaper in High Sierra that bit me good - install is proceeding now of VCSA on Fusion Pro 10.1.6 on a 2012 Mac Mini running High Sierra - this is gonna be interesting no doubt.
Installed VCSA successfully using the VCSA 8.0.1.00200 ISO on the old Fusion Pro 10.1.6 (the last version of Fusion that works on macOS High Sierra). All this running on the old 'high end' quad-core 2012 Mac Mini Server modified with 16GB RAM and a 500GB SSD. Olde Time fun it seems - 2 questions though - everything looks good so far, but I am getting an Alert in the VCSA UI saying memory is close to full (no doubt because it took 14 of my 16GB) - I've killed the vSAN service because I have no use for vSAN, but are there other things I can kill and maybe reduce by 1-2GB of memory? I will literally only have 3 ESXi hosts when fully deployed and I estimate a total of 30 guest VM's grand total in the environment. Second question is about storage - the "Tiny deployment" i chose mentions needing 579GB of disk - I ignored that and the install went fine. But keep in mind I only have a 500GB SSD in this Mini, so if it ever needs that I'm screwed with no upgrade path. So far it's only using about 50GB of space, what is going to use so much disk in VCSA in the future and can I do anything about that to prevent the Bad Day from coming? Thanks William et al.
See https://williamlam.com/2023/09/what-is-vc-ws1a-broker-service-on-vcenter-server-appliance-vcsa.html on how to save another 2GB of memory
Thanks William! I'll try that, should do the trick...any comment on why a tiny deployment needs 579GB of disk, when at least so far mine is only using 50-60GB. I'm sure there must be a simple explanation.