Tests if your nodes are getting slow in pod creation / termination.
k8s-node-check creates one (1) pod on every node and deletes it in a loop.
If the creation/termination takes too long, it marks the node as PIDPressure
status/taint that is then removed by kubelet if it's healthy.
It also checks for pods that are older than their node (kubernetes bug with spot instance replacement) and force deletes them.
$ k8s-node-check -create 10s -terminate 15s -every 5s -pods=10m
2022/12/10 20:00:11 PROBLEM CREATE ip-192-168-29-39.eu-north-1.compute.internal 11.438242s
2022/12/10 19:59:47 PROBLEM TERMINATING ip-192-168-75-31.eu-north-1.compute.internal 17.589388s
docker image at https://github.com/matti/k8s-node-check/pkgs/container/k8s-node-check