Hi there,
pretty old thread, but still the one which exactly deals with the problem we have today.
Situation:
4 node esx cluster, running on vsphere 4.1i update 3, build 800380, using vmware HA and vmware drs - fully automated.
vCenter running on version 5.0
Distributed vSwitch with 4 uplinks on each host, running as a 4.1.0 dvSwitch.
Today we tried to evacuate on of our hosts to maintain it tomorrow.
The vCenter scheduled all migrations and suddenly stopped doing anything. We found out, that it tried to migrate 4 vms to the same host (normally no problem), but this host had issues with the distributed switch.
Checking the hosts network settings seems to show normal values, but checking directly connected to the host with vsphere client, the host reported "This host has no physical adapters connected to distributed switch".
This message was weird, as the already running vms were all online.
With vmware support we had to stop vcenter service as it had a lot of issues configuring the dvswitch on the "problem" host. After restart of service at least vcenter was available again and showed, that all tasks concerning vm-migrations were cancelled.
After that we evacuated the problem host and removed all uplinks from the dvswitch. Reboot and connected the vmnics to the dvswitch again.
Still the host could not determine which of his vmnics are connected to the dvswitch. Therefore i removed all uplinks again, disconnected the host from vcenter, removed the host from dvswitch and rebooted it again.
After boot i connected the host again to vcenter. Now the host can determine which uplinks are connected, but it can't determine which portgroups should be availabe on this dvswitch!
I migrated a test-vm to this host and it remained online and reachable, but on the host i see in the vm-settings, checking the portgroup "Invalid Backing" - and i'm connected as root. In advanced settings i can see the correct dvswitch-port, but not the portgroup.
After 1 h the host got the right port-group showing up in network settings, dvswitch section. So i migrated some other vms to it, vms of other portgroups. But now, after finally 2 more hours the host still can't determine the new (and used) portgroups, but the vms are online.
Does someone has an idea on that?
Our vCenter does not see any problems, but i think all hosts should see their right network-settings. All other hosts in that cluster see correct data.
Attachements:
invalid_backing_host.jpg
a look on vm-settings directly connected to the problem esxi host
vcenter_information.jpg
a look on vm-settings on vcenter