I recently ran into situation that I can no longer vMotion VM from either my on-premise datacenter or VMC SDDC, previous it can do both vMotion and cold migration, but now I can only perform the clod migration but not vMotion. I got the error message showing something like the below screen captured.

failed vmotion

So I went to HCX GUI and try to fix that by checking all the tunnels for both the IX and NE between on-premise and VMC SDDC are all in “up” status. I found the NE appliance status was “down”, so I tried to “FORCE-SYNC” the IX and NE appliances between both sides, but strange enough I got a never ending job like below.

slack

My first think was to fix this by unextending the existing network extension and try to recreate it again, then I got the “Tearing down” status hanging there like forever.tearing down

I restarted on-premise HCX manager and all the HCX appliances on both sides, but still got the same result. You know that it’s not possible to “restart” or to “reboot” the HCX manager in the cloud side but to upgrade it, so I  decided to upgrade HCX manager on both sides to the latest R140 version, I know there will be a power cycling event after the HCX manager upgrade, that might help to clean up some unwanted hanging jobs.

uograde HCX

After the upgrade, I still saw the “tearing down” status in the “Network Extension” GUI, so I decided to use the “Force Unextend Network!” option to break the network extension “violently”.

P.S: You can use this option in your testing environment for a quick fix, but for production environment, please file an support request in your VMC console to get a root cause analysis first before you perform this task in order to prevent unwanted VM communication error.

 

  • In the vSphere Web Client navigator, select HCX.
  • Navigate to the Network Extension tab. You can see a list of currently extended port groups.
  • Locate the network that must be unextended and expand the selection.
  • Click Unextend.
  • Click the left arrow sign to expand extra available options.

 
tearing down 2

tick the “Force Unextend Network!” check box and click “UNEXTEND” button.


tearing down 3

You will see some updating jobs running in the vCenter GUI “Recent Tasks” windows.

progresses

After some time the hanging “Tearing down” items disappeared! Problem solved!

done tear

Then I can successfully recreate the network extension just like before.

recreate NE
extending

The extension can be successfully recreated again and showing the right status.complete

Now I can perform the vMotion tasks leveraging network extension capability again!

Migration succeeded

I hope this blog can help you deal with vMotion issues regarding network extension related problem, thank you!