ESXi VMkernel Interfaces "vmk#" vMotion network fallback to APIPA IP address when set to DHCP config Causing vMotion failure
search cancel

ESXi VMkernel Interfaces "vmk#" vMotion network fallback to APIPA IP address when set to DHCP config Causing vMotion failure

book

Article ID: 415837

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

  • DHCP server is unavailable.
  • One or more ESXi host has expired DHCP lease period.
  • APIPA IP is being assigned to "vmk#" vMotion VMKernel adapter.
  • APIPA IP reference range 169.254.0.0/16 to 169.254.255.255/16.
  • One or more host still has the previously assigned DHCP address as they lease is not expired yet. For e.g. 192.x.x.x/24.
  • "vmkping" fails between the hosts as the above will cause hosts to be in 2 different subnet.

Environment

VMWare vSphere ESXi.

Cause

vMotion vmkernel adapter is configured for DHCP, but its DHCP DISCOVER messages don't reach a functional DHCP server, or the DHCP server's OFFER doesn't reach the ESXi host, will result in an APIPA address being assigned

Resolution

  • Make sure the DHCP server is available on the network and reachable from the ESXi host to complete the D-O-R-A process to get the IP address from DHCP server.
    • The DORA process is the four-step method DHCP uses to assign IP addresses: Discover, Offer, Request, and Acknowledge. A client broadcasts a Discover message to find a server, the server replies with an Offer of an IP address, the client sends a Request to accept the offer, and the server sends an Acknowledge message to finalize the assignment.
  • If in case there is no DHCP server available in the environment, Make sure to use static IP address for the vMotion network.
  • It is recommended to use static IP addresses for management-related VMkernel adapters such as Management, vMotion, and vSAN etc.
  • To determine the communication issue, perform the packet capture using Packet capture on ESXi using the pktcap-uw tool

Additional Information

  • If vMotion with APIPA IPs being assigned to all the hosts in cluster using same portgroup will still work as they they can still reach over layer 2. 
  • Reference KB for the packet capture and analysis: Packet capture on ESXi using the pktcap-uw tool
  • reference KB for How to troubleshoot DHCP issues: 373392