Knowing what your Kubernetes cluster is doing
05-19, 11:35–12:05 (UTC), Main Hall

While running Kubernetes in production, how do you know what the cluster is doing? In this talk Federico will show you how he and his team are using kube-state-metrics in combination with other exporters and logs to get insights into the multi-tenant Kubernetes cluster they run for 40+ development teams at Meltwater. He will focus on metrics for the higher level Kubernetes objects as well as the cloud environment they run the cluster in.


This talk builds on and extends two other talks from previous KubeCon events:

  • Reveal Your Deepest Kubernetes Metrics (KubeCon 2018 Europe)
  • Monitor The World Meaningful Metrics for Kubernetes Applications and Clusters (KubeCon 2018 NA)

It will start where the other talks end and show real world examples of using metrics from kube-state-metrics for the higher abstraction level Kubernetes objects in combination with other exporters such as aws-limits and log files to get insights into the Kubernetes cluster we are running internally for other development teams at Meltwater. By running Kubernetes in production for over a year we have learnt (sometimes the hard way) to focus on the overall state of the cluster without loosing the possibility to dig into specific details as well as knowing about the state of the cloud environment the cluster runs in. The hard way gave us learnings on which metrics we were lacking and which ones were not needed.

Federico has both worked as a developer and infrastructure engineer - writing, building and running large scale systems. He has spent the last year developing and providing Kubernetes as a Service for 40+ teams internally within Meltwater. Federico is also an avid OSS enthusiast and has been a maintainer for a smaller OSS cli tool for a decade.