Upgrade Nutanix Community Edition To The Latest Versions

!!! Always make sure you have a working backup when performing upgrades !!!

The last year we had lots of new version in the Nutanix HCI ecosystem. Community Edition (CE) is built on the production version but had some issues with the latest versions. In this blog post I will show you to where you can upgrade (and their pitfalls). All screenshots and upgrades performed in this blog post are being done in an Intel NUC (i7, 64GB ram) with the CVM on a 1TB NVME and AHV on a 1TB SSD.

At the moment (August 2026) the latest CE version is 2.1 which hold AOS 6.8.1 and AHV 20230302.101026 when running LCM you can upgrade to:

• AOS 7.0.1.15
• AHV 10.0.1.9

This is still a “safe” upgrade path which you can do even if you have USB drives in your CE installation.

When you install to AHV 10.x or newer on consumer grade hardware (like a simple Intel NUC for example) you can’t run any virtual machines anymore. So when the installer upgraded AHV to 10.0.1.9 (for example) VM’s will not boot. To fix this run the script created by Kurt: https://github.com/ktelep/NTNX_Scripts/tree/main/CE/ahv10_commercial_workaround

curl -fsSL https://raw.githubusercontent.com/ktelep/NTNX_Scripts/main/CE/ahv10_commercial_workaround/fix_ahv10_comm.sh | bash

curl -fsSL https://raw.githubusercontent.com/ktelep/NTNX_Scripts/main/CE/ahv10_commercial_workaround/fix_ahv10_comm.sh | bash

When it is applied VM’s are working again.

Now we have a working Nutanix Community Edition on AOS 7.0.1.15 and AHV 10.0.1.9. (latest 7.0 and 10.0 releases at the time of writing)

To upgrade further, like AHV 10.3, there are USB limitations. In 10.3, and newer, the USB drivers are removed. So when you are running AHV (or CVM, DATA) from a USB drive you cant upgrade further. In that case you need to reinstall your cluster without any USB drives. An example on this you can find here: https://www.jeroentielen.nl/nutanix-community-edition-on-2-disks/

When there are no more USB drives in the cluster/node we can upgrade to AOS 7.3.1.10 and AHV 10.3.1.9 (latest 7.3 and 10.3 releases at the time of writing provided by LCM)

This AHV upgrade will be a long one. Follow it on the console to see the magic happening.

When the cluster is upgraded we run an LCM Inventory to see what versions are available to upgrade to:

At the moment we cannot go any further via normal LCM. But if you are a customer/partner of Nutanix you have access to the download you can download the AOS 7.6 and AHV 11.2 LCM binaries. When you have downloaded them upload then in LCM (direct upload) and do a inventory again:

Don’t upgrade yet, there are things changed in the AHV 11 release. If we just YOLO it now it will not work. We need to truncate our disk serials to a max of 20 characters.

Run this line of code in the AHV console:

grep -o '"serial": "[^"]*"' /etc/nutanix/config/cvm_config.json

grep -o '"serial": "[^"]*"' /etc/nutanix/config/cvm_config.json

This will show you the disk serials in use. In my case my Intel NUC is safe and I can run the upgrade. But let me startup a nested Nutanix CE installation on VMware ESX and run that command:

As you can see my disks have long serials (don’t count them, they are 32 characters each) and when we upgrade the cluster will not work anymore (yes the fixes can be done later as well, but better do it now). So first stop all running virtual machines on the cluster (including Prism Central). Then open ssh session to the CVM and stop your cluster and shutdown CVM.

cluster stop (i agree)
sudo shutdown -h now

cluster stop (i agree)
sudo shutdown -h now

When CVM is not runing open SSH session to AHV and run the following commands:

# First we make a copy of the original CVM configuration
cp /etc/nutanix/config/cvm_config.json /root/cvm_config.json.bak

# Truncate all serials to only 20 characters
sed -i -E 's/("serial": ".{20})[^"]*"/\1"/' /etc/nutanix/config/cvm_config.json

# Verify if the serials are 20 characters
grep -o '"serial": "[^"]*"' /etc/nutanix/config/cvm_config.json

# Recreate and start the CVM
systemctl restart ahv-define-cvm

# First we make a copy of the original CVM configuration
cp /etc/nutanix/config/cvm_config.json /root/cvm_config.json.bak

# Truncate all serials to only 20 characters
sed -i -E 's/("serial": ".{20})[^"]*"/\1"/' /etc/nutanix/config/cvm_config.json

# Verify if the serials are 20 characters
grep -o '"serial": "[^"]*"' /etc/nutanix/config/cvm_config.json

# Recreate and start the CVM
systemctl restart ahv-define-cvm

It looks like this:

When CVM is booted, SSH into the CM and wait for zookeeper to start (genesis status) and the give the following command:

edit-zeus --editor=vim

edit-zeus --editor=vim

When you are in the zeus editor search for the disks and edit their disk_serial_id and mount_path to 20 characters (remove the last bit). Make sure they match the serials in the cvm_config.json

Before:

After:

Save the file with “escape -> “:wq” –> enter

Verify the config in zeus with:

zeus_config_printer | grep -nE "disk_serial_id|mount_path|offline_disk_mount_paths"

zeus_config_printer | grep -nE "disk_serial_id|mount_path|offline_disk_mount_paths"

If the config looks correct you can reboot the CVM and then start the cluster:

# Restart the CVM
sudo reboot

# When CVM is back up and Zookeeper is running (genesis status) start the cluster:
cluster start

# Restart the CVM
sudo reboot

# When CVM is back up and Zookeeper is running (genesis status) start the cluster:
cluster start

When the cluster is up and running again upload the AHV 11.2 and AOS 7.6 binaries do an inventory and upgrade. Eventually you cluster looks like this:

And YES you can also install Prism Central 7.6. Make sure you install at least the SMALL version as that will allow you to enable the Ultimate trial.