Long story short we are swapping over to newer Nutanix chassis and replacing a few nodes. The previous tech had issues imaging the new chassis. However, in the prism hardware tab, the chassis is present (i think the previous tech was able to get it added somehow despite previous issues). I was able to go through foundation and get the new chassis nodes imaged and ready to be added back to the cluster, but the information remains in Prism and we cannot remove it. The new nodes show up when we select the expand cluster option, but the information (host ips, etc.) matches what is already in Prism, so it cannot be added. We are using a 5.x AOS and cannot upgrade.
We need a way to forcibly remove the stale information regarding the nodes that is still present in the cluster.
So far we have tried the following with support and are hoping someone has possibly seen something like this before.
We tried removing the nodes through Prism, but the 1st node we tried to remove is showing the meta data drives greyed out and it shows host being removed. It has been this way for 2 weeks now.
When issuing the “ncli host get-rm-status” command, I still see the node as stuck in “MARKED FOR REMOVAL BUT NOT DETACHABLE”
However, the task list output does not show a task labeled “kEnteringMaintenanceMode” as described in Article #000014161 and the problem host in the host list reflects “maintenance mode = false.”
When entering the ncli host edit id=<host.id> enable-maintenance-mode=true command, we receive the following: “Error node <node id> removal in progress. Cannot set node id <node id> in maintenance mode.”
When we tried to issue the “ncli host delete <host id> force=true” command it says the process of node removal is started, but nothing is reflected in the task list and Prism does not show any progress and the node remains as “MARKED FOR REMOVAL BUT NOT DETACHABLE.”
Please assist if you happen to have any experience with this.