AWS EKS Automode - Node metrics monitoring

[ftmazzone]

Hi All,

We enabled Dynatrace application monitoring on an AWS EKS auto mode cluster. The monitoring is working fine, but we are missing node metrics such as disk usage, CPU utilization, etc.

I've noticed that running a container on the node, we can find some node metrics such as the disk capacity and usage.

kubectl run -i --tty busybox --image=busybox
df -h

 
Is there a solution from Dynatrace that would allow us to collect the Bottlerocket node metrics without developing a custom solution?

Kr,

Florent

[engineerakki]

Looking for same, havent found anything.
It seems Dynatrace does not support EKS Automode properly.

[sujit_k_singh]

Hi @ftmazzone 

EKS Auto Mode uses Bottlerocket OS for nodes, which are fully AWS-managed. Unlike standard EKS node groups, you cannot install OneAgent as a DaemonSet on the host as Bottlerocket doesn't support traditional package installation. Without OneAgent running at the host level, Dynatrace cannot collect full infrastructure metrics for the underlying nodes.

The CloudWatch integration is probably your best bet for getting node metrics into Dynatrace without a custom solution.

Thanks,

Sujit

[ftmazzone]

Hello @sujit_k_singh ,

Exactly using the default setup is not working but running `df-h` on a container will provide some node metrics, this is an alternative solution to replace the daemonset that will provide some metrics. Node exporter seems to be compatible too.
Dynatrace should look in that direction to provide the Bottlerocket node metrics