Revamp Kubernetes Metrics
Our Kubernetes metrics are slow, painful, and do not show the entire picture of the situation that we may be dealing with. Example scenario where it's hard to determine failures:
- High rates of Pod restarts for a service
- High turnover of nodes
- Lacking saturation metrics for Pods and Nodes (and clusters for that matter)
- Some dashboards fail to load entirely
- Location filtering (targeting our regional or zonal clusters)
We utilize the k8s-mixin and are leveraging it's use inside of our already existing metrics platform. Utilize this issue to determine what we can do to make actionable/browsable dashboards useful for all teammembers.
Edited by John Skarbek