Recently I have discovered the kubectl describe node/ command, and noticed that some of my k8s nodes where not optimised for the workloads running on them (eg. <30% CPU limits, and 150% memory limits, basically overcommitting resources). This made me think that:

1. I might not have configured properly pod requests and limits

2. The k8s node instances might not be optimised for the workloads running on them

How do you track, manage and optimise resource allocation on k8s clusters (from pod limits to node types) ?