we have an esxi 5.5 host that is part of a cluster, and uses a dvswitch for networking. a couple of days ago, after upgrading the firmware on the server (to try and fix some other issues), whenever the host was rebooted, the management interface stopped responding. after doing some digging, the management interface had somehow switched over to an NFS portgroup, and the management option for the vmknic had been checked, so once we figured that out, did a direct connection to the host to uncheck it and restarted the services hoping that it would go back to the original adapter. marked it off as maybe an inadvertent click on the nfs vmknic. however, once the services were restarted, the host was still only accessible on the nfs vmknic, even though the management had been unchecked, and the original ip was still down, though it showed connected. did a reboot just to start over, and the same thing happened again. went through the steps here, http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=100812..., to move it back over to a standard switch, and it started working, even after reboot. after the reboot, tried to migrate the virtual adapter back over to the dvswitch using the host's section of vsphere, but if failed out with a "general system error occured: an existing connection was forcibly closed by the remote host" error. went to inventory/networking, and migrated it there no problem from the dvswitch/manage hosts page. but, same old story, after reboot, it reverted back to the nfs interface for management. so, this time, reverted it back to standard switch for management, remove the host from the dvswitch, rebooted, fine. added the host back to the dvswitch, but only added the vmknic for management, nothing in any other port group, rebooted, same thing, inaccessible, except this time it didnt have any other virtual adapter to use. we've got two other hosts with no problems with this. any ideas?
So i got it to survive a reboot by using the vds restore in the troubleshooting options, but it put the vmnic in a "hidden" port group, starting with hlp-. adding another adapter and removing the old in this port group resulted in lost connection to the host.
This may be a NIC driver or firmware issue.
What type of hardware (system vendor/model) and NICs are you using?
Have you configured VLANs?
And please try to format your posts in a more readable way. Thanks!
Andreas
it's a dell r620. this did happen after we ran an upgrade on the firmware. we're also using dell's vmware image since it has the drivers included.
Broadcom netxtreme II 57800 on board ethernet- no vlans.
We are an HP shop, so I do not have any experience with Dell servers.
But a general recommendation would be to update the NIC's firmware to the latest level and also the NIC's driver (bnx2x).
For the latter see
VMware vSphere 5: Private Cloud Computing, Server and Data Center Virtualization
Andreas
the firmware is the latest, as this is what seemed to cause the issue. i'll try the driver update. Thanks!
updating drivers didnt help.
You are using a vDS for Management? I have always use standard vswitches for management and vDS for the virtual machine vLANS. How many physcial nics do you have on that host?
