Solved

Move ESX nodes from 1 cluster to another

  • 8 January 2021
  • 2 replies
  • 220 views

Badge

Hey all -

We removed 2 nodes from 1 cluster and want to add them to another.  We disconnected from vCenter and then removed them from the cluster using PE.  When they were relocated to the new location, the vlan is the same and they could keep their same IP addresses, but the new cluster could not see them.  It sounds like the customer had configured Distributed switches on the old cluster.  Do I need to undo some of that networking to see the hosts?  Would bare metal foundation be a better option and put AHV on the nodes?  That is our next route, but I still think there is a problem with the network configured in ESX even for that.  Is there any documentation around undoing the vmware networking for bare metal?  With factory nodes, I would have just added them to the same vlan and assigned IPs using crashcart and then the cluster should have seen them.  

Help!

icon

Best answer by Alona 14 January 2021, 04:58

Hi Mellissa,

Let’s break it down a little:)

When they were relocated to the new location, the vlan is the same and they could keep their same IP addresses, but the new cluster could not see them.

Do you mean that other ESXi hosts of the new cluster could not reach (ping?) the hosts from the old cluster?

It sounds like the customer had configured Distributed switches on the old cluster.  Do I need to undo some of that networking to see the hosts?

Your main goal is to establish network connectivity between the hosts. The DvSwitch is managed by the VC. Ideally, you’d re-add the host back to the original VC, configure standard vSwitch with an uplink first otherwise you'll lose the connectivity to the host, remove the DvSwitch from the hosts. Then go and do what you set out to do.

Would bare metal foundation be a better option and put AHV on the nodes?  That is our next route, but I still think there is a problem with the network configured in ESX even for that.  Is there any documentation around undoing the vmware networking for bare metal? 

Alternatively to the previous point, you can do a Network reset from DCUI.

 

Some links you might find useful:

Host restored to factory network settings (2000538)

Delete distributed switch from host when vcenter is gone?
 

Hope this helps. Let me know how you go and if you have questions.

View original

This topic has been closed for comments

2 replies

Userlevel 6
Badge +5

Hi Mellissa,

Let’s break it down a little:)

When they were relocated to the new location, the vlan is the same and they could keep their same IP addresses, but the new cluster could not see them.

Do you mean that other ESXi hosts of the new cluster could not reach (ping?) the hosts from the old cluster?

It sounds like the customer had configured Distributed switches on the old cluster.  Do I need to undo some of that networking to see the hosts?

Your main goal is to establish network connectivity between the hosts. The DvSwitch is managed by the VC. Ideally, you’d re-add the host back to the original VC, configure standard vSwitch with an uplink first otherwise you'll lose the connectivity to the host, remove the DvSwitch from the hosts. Then go and do what you set out to do.

Would bare metal foundation be a better option and put AHV on the nodes?  That is our next route, but I still think there is a problem with the network configured in ESX even for that.  Is there any documentation around undoing the vmware networking for bare metal? 

Alternatively to the previous point, you can do a Network reset from DCUI.

 

Some links you might find useful:

Host restored to factory network settings (2000538)

Delete distributed switch from host when vcenter is gone?
 

Hope this helps. Let me know how you go and if you have questions.

Badge

Alona -

We bare metal imaged the nodes to AHV with the correct IPs (not forming a cluster) and everything worked fine.  One issue I saw was the foundation process was not running after these were removed from the old cluster.  I was able to walk the customer through the foundation process using the Windows stand alone software and everything was good to go after that.  We could do pings with no issues when the nodes had ESX on them, but without foundation running, I’m guessing they wouldn’t be discovered by the new cluster.

Thanks!