Since release 283 we are not propagating host names to Kubernetes nodes anymore, but only use the Kubernetes node name as provided by the Kubernetes API.
Why?
The node name in Dynatrace now reflects the node name as it is seen with the
kubectl get nodes
command. This makes it easier to correlate data from the API with the data visible in Dynatrace.
What does this mean for historic node entities?
Kubernetes nodes, which have been first touched before release 283, internally might still hold the inherited host name. This inherited host name will not be deleted. As it overrules the real node name, when it comes to compute the entity name, it will still be visible in the UI.
Are there affected Kubernetes node entities?
The following DQL query lists all kubernetes node on the tenant sorted by the timestamp when they were first seen.
Less recent nodes might (but not neccessarily need to) have differences between the detected name (as seen in the kubectl command) and the entity name (which might be the inherited host name).
fetch dt.entity.kubernetes_node
| fieldsAdd lifetime, entity.detected_name, clustered_by, entity.detected_name == entity.name
| sort lifetime DESC
