Hi Elastic team,
I wanted to share some feedback and ask about an issue I’ve observed with the Kubernetes Prebuilt Cluster Overview Dashboard coming from the Kubernetes Integration. The visualization for "Memory used vs total memory" relies on the field kubernetes.container.memory.usage.bytes for memory used, but in my environment (and others I've seen), this field is always null or zero. However, in the same log files, the field kubernetes.container.memory.workingset.bytes is populated with meaningful values.
From what I understand, this seems to be related to changes in Kubernetes (v1.20+), where the working set metric is now prioritized over the older usage metric for representing actual memory pressure. The working set is generally considered a more accurate measure of memory that is actively in use and cannot be easily reclaimed.
Because the dashboard is using the legacy field, the visualization doesn’t work out of the box in modern clusters. I’m wondering if Elastic plans to update the integration and dashboards to use kubernetes.container.memory.workingset.bytes instead, or if users are expected to manually adjust the visualizations themselves. I’ve noticed this isn’t just an issue in my environment, so it might be affecting others as well.
Also, I’d like to mention that setting up the Kubernetes integration can be quite complex compared to some other monitoring solutions, which makes these kinds of issues a bit more challenging to troubleshoot.