Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This would break my cluster. I use non-existent packs on purpose for ad-hoc rules and policies that a workflow I wrote creates when deploying an updated pack. So, they're very likely to exist when upgrading st2.

As a convention, I use dunder names (__some_pack__) for packs that are not expected to exist on disk. I use a policies to delay all of the actions in a pack during a pack upgrade. I'm planning to use ad-hoc rules for scheduling the run of some action at a later time.

I do agree that this is a problem. But, I suspect we need more primitives within st2 core to handle this in a cleaner way, and in a way that is not k8s-specific.