Kubernetes Metrics Server memory usage reporting

Posted on March 14, 2024

I have installed Kubernetes Metrics Server on my cluster using a ‘1-Click App’ from the ‘Marketplace’ tab on the DO website. Now, when I connect to my cluster using K9s, it reports a memory usage of 94%. But when I go to the ‘Insights’ tab for my cluster on the DO website, I get reported a memory usage of about 54%. Does anyone have an idea where this (big) difference comes from? (The reported CPU usages do seem to match.)

This textbox defaults to using Markdown to format your answer.

You can type !ref in this text area to quickly search our full set of tutorials, documentation & marketplace offerings and insert the link!

These answers are provided by our Community. If you find them useful, show some love by clicking the heart. If you run into issues leave a comment, or add your own answer to help others.

Bobby

March 17, 2024

Hey!

I’ve also noticed a similar behaviour with K9s for some of my clusters that I’m running onprem.

I think that it’s because K9s typically only has a node-level perspective, eg. memory usage from the viewpoint of an individual node in the cluster.

But on the other side the DigitalOcean Insights has a cluster-wide perspective and aggregates memory usage across all nodes and underlying infrastructure components rather than a single node.

Also, the way metrics are aggregated and averaged over time can slightly influence the reported values.

I believe that with any tool, it’s normal and expected to see some difference, and what you can do is to rather than fixating on precise percentages, observe the trends reported by both tools. Are memory usage spikes in Metrics Server mirrored in DigitalOcean’s Insights? This is often more helpful for operational insights.

Hope that this helps!

Best,

Bobby