I've been working on an issue with VMWare for quite a while, and thought I would engage the community for input as well. We have a problem where seemingly random ESX 4.0 and ESXi 4.1 hosts will lose the physical adapter dvUplinks on the vNetwork Distributed Switches after patch remediation using Update Manager. The issue is easily fixed by reappling the host profile to the host, however that is an extra step for our patching group and we're running into permission issues with that work-around.
This week, VMWare noted an internal KB article that was not directly related to our issue, but allowed for the following theory (from VMWare.) Somehow an affected host gets "out of sync" with VC with regards to the vDS configuration and when it is rebooted (patching is just when we normally reboot the hosts) it is unable to get the proper vDS configuration from VC so the switch has no physical adapters assigned to dvUplinks. I'm not sure if I should be posting internal KBs but the essentials were that an ESX host was abruptly/improperly shutdown and the vDS cache located on the host was corrupted. Resolution was restarting the service agents on the host.
Has anyone else ever run into this issue? Is there some sort of PowerCLI script that will help me diagnose "out of sync" hosts in our environment? Any thoughts or ideas on how to continue troubleshooting would be appreciated.
A few details:
ESX 4.0 builds 702116 and 721907
ESXi 4.1 builds 702113 and 721871
Virtual Center Server 5.0.0 build 623373
Thank you,
Tim
