Disable Oneagent Monitoring for Kubernetes Cluster

[agylpradipta]

Hello All,

Currently I'm testing Dynatrace on my dev environment.

I want to temporary disable the oneagent monitoring for it so it will not consume license.

Is there any way to disable monitoring on node of kubernetes cluster instead of manually disable it one by one?

Thank you in advance.

[Mizső · Accepted solution]

Hi @agylpradipta,

I thnik you should try the combination of Monitored entities and Settings objects API.

Enable/Disable by settings objects API:

I hope it helps.

Best regards,

Mizső

[Gabor]

Hi Mizső,

Is this the same as if I disable it in the kubernetes menu, when I select a cluster and in the monitoring settings I disable the monitoring?

My problem is I have to somehow disable monitoring to not only save DDUs but free up Host units, without having access to AWS where our AWS EKS clusters are running. Is there any possibility to do that via the Dynatrace UI or Dynatrace API?

[Gabor]

Update:

Host units are not freed up with disabling monitoring this way.

The solution for "dynamic" host unit license management on cloud is the container monitoring rules.

Container monitoring rules has to be created based on e.g. k8s namespaces this way it is possible to disable montoring on different / not needed namespaces saving host unit consumption without redeploying pods / operators etc. 

[Mizső]

Hi @Gabor,

As far as I know container monitoring rules works only with FullStuckClassic operator mode.

Best regards,

János

[Gabor]

Hi @Mizső ,

It is possible, We havent tried it out yet, I base my answer on a tip we recieved.

In the following months I will implement these rules for all of our clusters for all of our namespaces and do some testing with license consumption and get back with my results.

Best regards,
Gabor

[Michel-Vallee]

Hello Gabor,

On our side, we are currently evaluating different options to suspend Dynatrace Kubernetes monitoring on our development clusters during evenings and weekends.

I would like to know if you have made any progress following your last message?

More specifically, I am looking for a solution that would allow us to centrally control this suspension (Dynatrace SaaS), ideally from the perspective of container monitoring rather than host (node) monitoring.

This is a temporary measure, as our long‑term objective is to partially shut down most of the infrastructure supporting these clusters. This would help reduce not only Dynatrace monitoring costs, but also overall infrastructure costs.

Best regards,
Michel

[Julius_Loman]

Hello @Michel-Vallee ,

I've implemented such a solution for one of our customers. It's based on Workflows in SaaS and does exactly what you describe. Parts of the environment (typically non-prod environments) are disabled and enabled on purpose.
Triggers are:

• schedule (enabled 6:30 - 19:00 on business days)
• manual
• when deployment is running (integration with gitlab and SDLC events), so that you don't need to restart OneAgents.

After enabling, the workflow checks processes where deep monitoring is not working and sends events that a particular process has to be restarted - this typically does not happen, but sometimes people do manual changes instead of deployment.

It not only includes OneAgents, but also Kubernetes clusters and can be extended to any other configuration. This makes perfect sense in combination with DPS.

[Theodore_x86]

This is a quite important but I am not sure if it is feasible.

You can of course delete the dynakube file each time you want to disable OneAgents and recreate it but this is a process executed on the cluster and not on the Dynatrace.

Maybe this could be a new Idea. To add a Monitoring State button for K8S/Openshift clusters.

What do you think?

BR