In a VMware vSphere 5.5 environment we discovered a strange issue with most VM’s. After a reboot they would lose their network connectivity. It doen’t matter if they run 2008 R2, 2012 or 2012 R2, they all show the same behavior.
- They are unreachable over the network
- In vSphere they show the correct IP address
- When logged in on the console of the machine, it shows an APIPA address next to the fixed IP address
- Making any change to the VM NIC triggers the VM to use the correct IP again. This includes the ‘connect at power on’ option.
When trying to reproduce this we noticed a consistent behavior.
- The first OS initiated reboot goes without any problems
- If a second reboot is initiated, the machine won’t have network connectivity. So all updates with a double reboot are a pottential issue.
After some fact finding with our search engine of choice, we discovered the following thread on Technet.
After downloading and installing Xarp we found that we indeed got alerts.
Every now and then an invalid IP address would be detected: 0.0.0.0 with the MAC address of one of the Cisco switches.
We ended up solving this by adding the following to the config of the trunks on the Cisco
ip device track max 0
After this no more alerts were generated in Xarp and we were able to reboot a test VM several times in succession without losing network.