I have a GitLab Runner stack that has a couple of sidekicks that massage the config nad get the Runner registered. The sidekicks are labelled as “io.rancher.container.start_once: ‘true’”. It works well under normal circumstances. However, if I take down and then bring up the whole cluster, including the underlying Docker Engines, it looks like Rancher starts the “start_once” sidekicks again.
Is this normal behavior?
Should the “started once” persist over the cluster restart or should it not?
Thank you!