Disconnect/ connect NIC results in Invalid configuration for device '0'.

[1337Thomas]

Hi,

anybody having the same experience? Or any ideas where to look for the solution?

environment:

working infrastructure. ESXi 6.7 latest build vcenter latest build

hosts are on HCL and running latest firmware.

dvs is running version 6.5

VM tried different OSes (win and linux)

HW 15 and running latest tools.

HW 13 latest tools and old tools

HW 13 and os independent with linux

scenario:

you disconnect a nic and reconnect is and receive error: Invalid configuration for device '0'.http://www.joshodgers.com/2014/12/08/virtual-machines-lose-network-connectivity-after-ha-fail-over-e...

power on after you disconnected the nic does not connect the NIC on boot

rebooting a working server does not disconnect.

the workaround is putting the NIC in a different portgroup and putting it back in the original PG. when changing PG you don't have to connect it to work after changing it back.

looks a bit like this one:

Virtual machines lose network connectivity after HA fail over. (ESXi 5.5) | CloudXC

[andvm]

Had the same issue when I imported/created VM from a template which was connected to a port group which does not exist in the destination host.

I removed the nic and re-added

[1337Thomas]

We have it on the entire cluster. Not all are made from the same template.

Also no difference in e1000 of vmxnet3.

Adding a second nic to the non-function VM in the same portgroup works without issues. Until you disconnect it and reconnect it.

[chriscooney1982]

We have the same issue, we logged a support ticket with VMware who are aware of this as a known bug in ESXi 6.7 U3 with no fix yet.  Engineering team is working on a fix and will publish a KB article soon.  In the meantime the workarounds are:

1  Use the flex web client to reconnect.

2. Reconfigure the VM on another portgroup, set to OK and then reconfigure back to original portgroup.

3. Use portgroup with ephemeral port binding, we haven't seen this type of portgroup being affected, however, this will require to move VMs there as in the point above.

4. Standard vSwitch portgroups are not affected

We use static port binding so it makes sense in our case.  These workarounds aren't ideal I know but it's proving to be a big issue for us as well so fingers crossed there is a fix soon!

[1337Thomas]

thanks. This saves a lot of time troubleshooting.

Could you please update the thread once vmware has a solution or a hotfix?

it seems that VMs connected to NSX-T (nvds) are also not affected

[chriscooney1982]

VMware support have closed the ticket now and have said they won't inform us directly when a fix is released.  Kind of up to us to monitor the VMware website I guess.  I will bookmark this page though and update when I see something.

[ninjabrum]

We are also experiencing the same issue when disconnecting / reconnecting vNIC on VM's hosted on ESXi 6.7u3 hosts.  Here's the VMware KB on this:

VMware Knowledge Base

Workarounds:

Use anything but the H5 client to reconnect the vNIC (Flash/Flex client, powershell, etc)

Switch the vNIC portgroup to any other portgroup, enable the vNIC, then switch back to the original portgroup

[nateeric]

I'm also experiencing this issue.  Seems to be a bit of confusion, so I'm not sure what steps to take here...I had a VMWare helpdesk person tell me this morning that the issue is a vCenter issue and is fixed in 6.7 U3g.  The KB posted above states the issue is resolved in ESXi vSphere 6.7 PO2, which is the latest ESXi patch.  Neither of the changelog documents mention this issue as a resolved issue in the list of fixes.  Anyone update to either of these yet and successfully fixed this issue?

[ninjabrum]

Confirmed this issue is resolved in vCenter 6.7u3g - just upgraded my lab to confirm.  The KB doc needs to be updated.

[nateeric]

Odd they don't mention it as a fixed issue in the release notes.  I'll get the update on the schedule for this weekend then. Thanks!