The basics of a Helm hook

A hook executes as a one-time action at a designated point in time during the life span of a release. A hook is implemented as a Kubernetes resource and, more specifically, within a container. While the majority of workloads within Kubernetes are designed to be long-living processes, such as an application serving API requests, hooks are made up of a single task or set of tasks that return 0 to indicate success or non-0 to indicate a failure.

The options that are typically used in a Kubernetes environment for creating short-lived tasks are a bare pod or a job. A bare pod is a pod that runs until completion and then terminates but will not be rescheduled if the underlying node fails. A bare pod differentiates from a standard pod by toggling the restartPolicy property. By default, this field is configured as Always, meaning that the pod will be restarted if it completes (either due to success or failure). Even though there are use cases for running bare pods...