All of a sudden, some hosts are being duplicated for some reason. In the UMP Inventory view, you can see many hosts that are duplicates. In the CM_COMPUTER_SYSTEM table, we noticed that there is one properly populated entry and it contains the mac address of the host, but for the dupes, the dedicated field for the dupes shows 'NutanixVM' and the rest of the fields are NULL. Cross-reference case 32457455.
Release : 20.1
Component : UIM - NUTANIX
Customer followed the KB Article-> "Duplicate, missing, inactive and/or broken robots in IM after upgrading hub to 9.31"...
...as they had disappearing robots, and a lot of duplicates, but there were some slight differences in this scenario from other recent cases of duplicate hosts occurring (Dedicated field showing 'NutanixVM' for ALL dupes).
The customer ran through that process 3 days prior, but a large portion of dupes resurfaced.
So, we basically ended up doing a 'discovery reset' using USM Inventory (since USM was still working in the customer's environment) by manually selecting and deleting any/all DUPLICATE Host entries containing 'NutanixVM' in the dedicated field that also did NOT have full discovery data populated, e.g., OS information, etc. Note that ALL invalid host entries were associated ONLY with the Non Prod origin, or were displaying the host in UPPERCASE or were decommissioned.
The duplicate host issue reoccurred on the following day with the same entries containing the dedicated field value set to NutanixVM.
But then, after a few days more the issue disappeared and then did not resurface.
This is what worked to finally remove all of the duplicates and prevent them from resurfacing:
We essentially performed a 'discovery reset' using USM Inventory (since USM was still working in the customer's environment) by manually selecting and deleting any/all DUPLICATE Host entries containing 'NutanixVM' in the dedicated field that also did NOT have full discovery data populated, e.g., OS information, etc. All invalid entries were associated ONLY with the NonProd origin, or were displaying the host in UPPERCASE, or were decommissioned, e.g., robots.
We selected 'Prevent Rediscovery' but left 'delete alarms' and 'delete QOS' unchecked.
Suspiciously, although ALL of those dupe entries were also associated with the Non-Prod Hub origin, we confirmed that the PROD hub is truly NOT able to contact the NON-PROD over the network. (To test it out, we tried rediscovering based on a single IP, and a) it did not rediscover/reestablish the dupe host entry and b) it did not delete/alter the existing valid host data-entry.
Therefore, now the Inventory entries are all cleaned up and valid but it took a few days.