VMware Networking Community
Marino_S
Contributor
Contributor

Remove old host information from NSX T Manager

Hi,

 

My customer is in the process of upgrade his vSphere 6 (yes, i know....) to vSphere 7.  

We are building a greenfield environment, so that we can resolve any issues that pop up,   during this work,  I was a bit too trigger happy and I removed an nsx enabled host from vCenter before I had removed it from the vsphere cluster, and from the distributed vswitch.

We are using NSX T for endpoint security

I had to reinstall that host, so it is added to vcenter with the same host name and ip address as it had originally.

I can add the host without any problems to vCenter but when I move it back in the cluster,  NSX T manager gives as validation error.  It states: "26210: Node <hostname><id> with <ip> already exists

I suppose somewhere in a database there is still a reference to the previous configuration,  that causes a conflict.  Is there a procedure to manually remove to old reference from NSX Manager.  I've been going through the doucmentation but can't find a solution.

VMware support is also not a very big help.

Any advice is much appreciated

Kind regards

Marino

 

Reply
0 Kudos
8 Replies
JaSo2
Enthusiast
Enthusiast

Something similar happened to me in my lab - I have found the host among Standalone hosts in NSX Manager GUI and was able to Force delete it that way. After that It was possible to add the old-new host under the same name / IP etc.

Reply
0 Kudos
Marino_S
Contributor
Contributor

Hi,

Thank you for your reply,  I saw it only today.

I had at ticket with VMware  support, and it took us quite some to resolve it,  it needed quite some calls through the rest api to fix this.

The lesson learned is , do not remove a host from inventory without verifying first that you have removed if from nsx.

Problem solved,  thank you

Kind regards 🙂

Reply
0 Kudos
M4nch0t
Contributor
Contributor

Hello,

I am facing the same problem on my lab, can you give the api calls you have made?

Thanks

Reply
0 Kudos
Tuco1
Contributor
Contributor

Hi 

 

These were the steps we took

 

1) We checked the compute manager connection and cm-inventory to see if we have any connection issues with VSphere and we found no error in connection.

2) We moved the host out of the cluster (in maintenance mode) and tried multiple api calls to delete the TN which did not work and the error persists .

Delete  /policy/api/v1/infra/sites/default/enforcement-points/default/host-transport-nodes/<transport-node-id> 

DELETE  /api/v1/transport-nodes/<transport-node-id>?force=true&unprepare_host=false 
(in the header of postman)
key : X-Allow-Overwrite
Value : true

3) We created a new empty cluster and put the host in it (in maintenance mode) and then we did the following command 

DELETE  /api/v1/transport-nodes/<transport-node-id>?force=true&unprepare_host=false 
(in the header of postman)
key : X-Allow-Overwrite
Value : true

After that we put the host back in the original associated cluster to let it re-configure with NSX again which worked.


good luck,

Kind regards,

Marino

AdamKithcart
Contributor
Contributor

Good evening, I am having the exact same issue as this. How were you able to get the UUID? When I run the API call I get back "Invalid UUID string when I use the <host><id><ID> string that the error throws. Assuming this is not the actual UUID it is expecting. Any help would be greatly appreciated!!

Reply
0 Kudos
degvm
Enthusiast
Enthusiast

The Error Message in NSX Manager shows the ID:

It states: "26210: Node <hostname><id> with <ip> already exists

Copy past hostname id to the api call.

Reply
0 Kudos
JaSo2
Enthusiast
Enthusiast

This is the UUID you are looking for - click the three dots next to host transport node:

JaSo2_0-1686922035980.png

Or simply run API call /policy/api/v1/transport-nodes/ and this way you get a list of all nodes.

For me worked DELETE /api/v1/transport-nodes/<transport-node-id> + identifying the hosts in Standalone hosts in Orphaned state, than I have force deleted them (it takes some time).

J.

Reply
0 Kudos
OctavianTavi_P
Contributor
Contributor

this also worked for me / one note: you need to wait a couple of minutes in order the changes to expire from some cache 

Reply
0 Kudos