I am experiencing that pods are scheduled to nodes which have reached their --max-pods. The pods fail with status OutOfpods. There is a lot of other nodes available, which have space for more pods.

Has anyone experienced that pods get scheduled like that?

I tried reading the documentation for --enforce-node-allocatable, but it really isen't that clear. I noticed that our kubelets are running with --enforce-node-allocatable="", does that have any influence on this?