GCP and GCP Kubernetes with check_mk where is the love?

Feedback
1

Hi there!

Since we updates our cluster to +1.8 our Check_MK 1.6p16 fails to read it. Maybe with p22 this is fixed, I will see.
But I tested with 2.0.0p3 to see what it will be with 2.0. So in short: no different / not much to see.
I still don’t get my usage inside the deployment, I only see Limits and Requests - so not much help to monitor my cluster in check_mk.

I really really hope to see some improvement here in the near near future.

And beside Kubernetes? I don’t see any integration whatsoever :frowning: E.g. no way to monitor my SQL in GCP with check_mk, no way to monitor my cloud storage.

Am I missing something?

And a special question regarding Kubernetes: I have some namespaces I don’t want to monitor - I would love to get filter.
On top of that, I don’t want to monitor some deployments. There is a regex to define what deployments I want monitor, BUT I cant negate it. “Monitor all, but not…”

2

Hey morbloe,

thanks for the feedback. More love for Kubernetes will happen, don’t worry. I am currently working on this with the team and this is exactly one of the topics, we want to address.

We have focused so far on AWS and Azure monitoring, but will extend both + add support for GCP this year as well.

We plan to add the namespaces as labels, then you could use them for filtering. And your use case with: “I want to monitor all, but x namespace” makes perfect sense. We will highly likely also include that.

In the meantime, I recommend using our Prometheus integration. This way you can get usage data as well.

Martin

3

could you please explain with more details, what I need to do and what I could accomplish with this approach, thanks!

4

You can use Prometheus as a data source for Kubernetes monitoring. We integrated relevant exporters, e.g. cAdvisor and kube-state-metrics. First one gives you the usage data for containers and pods. Latter one is similar to the current Kubernetes monitoring and gives you config info.
My colleague Sudhir did a demo of it at FOSDEM: Integrating Prometheus to Checkmk | Demo @ FOSDEM 2021 - YouTube
There is also an article in the documentation, which describes the steps: Integrating Prometheus

You however need a Prometheus instance running on each of your clusters, which you want to monitor.