The vCenter Server vpxd service stops within minutes of starting, causing the vCenter to show objects and items being unavailable. Only the vpxd service stops, and is easily restarted. However, it soon stops again. This issue occurs after reverting a vCenter Server to a previous version snapshot.
- VMware vSphere 8.0 and later
- vCenter Server Appliance (VCSA) reverted from 8.0 U3 to 8.0 U2
- At least some ESXi hosts managed by the vCenter running ESXi version 8.0 U3
This issue occurs when a vCenter Server running version 8.0 Update 3 is reverted to version 8.0 Update 2, and some ESXi hosts in the environment have already been upgraded to 8.0 Update 3. The older vCenter version (8.0 U2) is not aware of Embedded vSphere Cluster Services (vCLS) VMs introduced in Update 3, leading to conflicts on the upgraded hosts that cause the vpxd service to stop repeatedly.
/opt/vmware/vpostgres/current/bin/psql -d VCDB -U postgres
select id, dns_name, product_version from vpx_host;
\q
vim-cmd vmsvc/getallvms | grep vCLS
vim-cmd vmsvc/power.off <VM_ID>
vim-cmd vmsvc/unregister <VM_ID>
service-control --stop vmware-vpxd
service-control --start vmware-vpxd