You see multiple entities in the Explorer menu with the same IP, but different names.
This can happen when the device using that IP has been renamed, or the IP has moved to a different device. Your
ktranslate snmp-base.yaml file will have entries for every combination of
device_name__ip it has ever discovered. This is common when an IP is used for redundancy like with HSRP, switch stacks, or primary/secondary failovers.
When a discovery job runs, each time it discovers an IP it tries to determine the
device_name connected to the IP address. If there is not already an entry in the devices list that matches that specific combination of
device_name__ip then it will add a new entry with a hard coded
device_name. It does not automatically over write or remove previous entries that used that IP.
ktranslate performs its polling it only uses the IP when it runs, so if your devices list contains multiple entries with the same IP address, it will collect and send metrics to New Relic as if they were separate entities, but in reality it is just the same data that came from whatever device responded to requests to IP at the current polling interval. The
device_name is not collected or updated as part of the polling cycle.
device_name has changed due to a one-time change, like replacing a piece of hardware or updating your naming conventions, then you should edit the
snmb-base.yaml and delete the entry with the old device name. The old entity will still show in the Explorer menu but will stop having new data associated with it and eventually it will age out of the system. In most cases that will happen 8 days after data stops coming in.
Any IP that is part of a redundancy group or will be moved between different devices should be added to the ignore_list in the snmp-base.yaml. These devices should be monitored individually with a unique IP address that will reliably connect to one specific device.
If you want to confirm that the virtual IP is up you can add it manually as a ping_only device.