Hosts booted via autodeploy gets the Management IP as 0.0.0.0

Products

• VMware vCenter Server

Issue/Introduction

Following an upgrade to vCenter Server 8.0.3, diskless ESXi 7.0 hosts may fail to complete the Auto Deploy boot sequence. While the hosts successfully download the ESXi image, they may drop their management IP to 0.0.0.0 and enter Maintenance Mode, failing to apply Host Profiles or connect to the vCenter inventory.

The following log entries recorded during the boot sequence confirm a network health check failure and subsequent rollback

hostd.log

YYYY-MM-DDTHH:MM:SS info hostd[2104256] [Originator@6876 sub=Vimsvc.ha-eventmgr] Event 138 : Teaming configuration in the vSphere Distributed Switch on host esxi.example.com does not match the physical switch configuration in ha-datacenter. Detail: No loadbalance_ip teaming policy matches

vmkernel.log

YYYY-MM-DDTHH:MM:SS cpu37:2098125)Tcpip: 1605: Cleaning 3 Leaked routes on 'vmk0'
YYYY-MM-DDTHH:MM:SS cpu37:2098125)Tcpip: 1612: Freeing interface 'vmk0'

Cause

The issue is caused by a legacy Host Profile applying a "Route based on IP hash" (loadbalance_ip) teaming policy to vSphere Distributed Switch (VDS) uplinks when the physical switch ports are not configured for a Static EtherChannel. During boot, the ESXi network health check fails, triggering the "Network Rollback" safety mechanism which deletes the vmk0 interface to prevent a network partition

Resolution

Change the NIC Teaming policy on the Host Profile or change the NIC Teaming policy on the reference host and extract the Host Profile

1. Create dedicated vSphere Distributed Switch (VDS) Port Groups for Management, NFS, and vMotion utilizing the default "Route based on originating virtual port" teaming policy.
2. Manually connect a reference host to these new port groups.
3. Extract a native vCenter 8.0 Host Profile from the reference host.
4. Assign the new native Host Profile to the Auto Deploy rule and reboot the affected hosts