Following an upgrade of ESXi hosts (e.g., from version 8.0 U3d to 8.0 U3g) using a vSphere Lifecycle Manager (vLCM) image or after a remove host workflow in VMware Cloud Foundation (VCF), ESXi hosts display the following warning in the vSphere Client: "This host has no isolation addresses defined as required by vSphere HA"
Upon inspection, the advanced vSphere HA parameter das.isolationaddress0 is missing from the cluster configuration, and only das.usedefaultisolationaddress = false is retained. This occurs even if no manual changes were made to the isolation addresses prior to the upgrade.
VMware Cloud Foundation 5.x
VMware vSphere ESXi 8.0
vSAN (L2-based networks)
This is a confirmed defect within SDDC Manager. During a remove host or vLCM upgrade workflow, the system incorrectly triggers the UnConfigureHaIsolationAddressOptionsAction. This occurs because the "L3 Aware" component was removed from the underlying recipe.
Because the script fails to check if the cluster is L2-based—where all hosts share the same vSAN gateway and rely on that IP—it erroneously deletes the only HA isolation address. As a result, vSphere HA immediately alerts that the host has no isolation addresses defined.
This issue is resolved in SDDC Manager version 9.1.
Workaround: If an immediate upgrade to SDDC Manager 9.1 is not possible, implement the following to restore HA functionality and prevent recurrence during future workflows:
Log in to the vSphere Client.
Navigate to the affected vSAN Cluster > Configure > vSphere Availability > Advanced Options.
Click Add and re-enter the lost parameter: das.isolationaddress0 with the correct vSAN gateway IP.
Click OK, then right-click the affected hosts and select Reconfigure for vSphere HA.
To prevent SDDC Manager from erroneously stripping the address again before the environment is upgraded, SSH to the SDDC Manager appliance as root.
Append the following value to /etc/vmware/vcf/domainmanager/application.properties: configuration.disable.setting.ha.isolation.addresses=true
Restart the service: systemctl restart domainmanager