VM cannot be contacted following vMotion

Information

Title: 

VM cannot be contacted following vMotion

Author(s): 

Xtravirt (Gavin Jolliffe, Paul Davey)

Target Audience: 

Technical - Novice

First Published: 

1 April 2008

Products: 

VMware ESX 3.x

UID: 

XD10069

VMware ESX hosted VM cannot be contacted following a successful vMotion to another ESX host. This Hot Tip provides a possible solution.

1.0 Overview

Whilst working on a customer site, we witnessed first hand the problems that are resolved in a recent VMware ESX patch update. Following a successful vmotion activity on a VM, we discovered an intermittant problem whereby the VM could not be contacted via ping. However, using the console to access the VM and then running ipconfig /renew, or pinging a network resource, resolved the issue.

 

2.0 Solution

The problem is caused by the ARP cache on the switch not being updated correctly. As discussed in the VMware KB article that is associated with the patch, our network switches were connected in a stack configuration. Applying the patch resolved the issues and we experienced no more problems.

Note: You will not see this problem if you cold migrate a VM. When the VM comes back online and requests an IP address from your DHCP server, the ARP Cache on the switch will be updated.

We experienced the problem on an ESX 3.0.1 estate. The patch that resolves this issue was released in the November patch release. Patch number 2438 superceeds this patch, however, we only needed to apply the 2093 patch to resolve the issue.

References

References: 

1. Nil

Links: 

1. http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1002093