This tool is intended to help emulate the concept of "critical containers".
That means that when a critical container inside a pod stops, this code will see that (after a timeout) and if the container doesn't restart, the pod will be deleted.
This tool requires that the pod be annotated with
podwatcher/critical-containers: "container1" where the value is a comma
seperated list.
When there are multiple containers that are in the critical container list,
you can choose if any or all containers need to have terminated. You can
use podwatcher/condition to configure it. By default, any will be used.
This project uses Kustomize for deployment, it's recommended to update the
namespace. It will default to using pod-watcher-system, if you don't change
it you'll need to create the namespace.
A ClusterRole and ClusterRoleBinding will be created to allow the created ServiceAccount to access Pod and Jobs.s