hi everyone. A rather strange question - what do you think about implementing a cascading resource deletion for

pulumi state delete

? Let's say I royally screwed up and ended up deleting some resource that has a lot of children, e.g. a k8s cluster. Now my Pulumi state is horribly broken as

pulumi state delete

isn't really helpful since there are far too many dependencies to walk the tree myself from down up. In this situation it would be really great to have

--yes-i-know-what-i-am-doing

flag to

pulumi state delete

to force the deletes to cascade. After all, my root resource is not there anymore and the resources in the state do not correlate with the actual infrastructure objects anyway. wdyt?

At first glance, a cascading delete would be very risky, but a protect-if-has-children opt would be very safe. And maybe some way to get Pulumi to generate a script full of all the

pulumi state delete

commands necessary to fully clean up, so that I can manually edit it myself..

Such a flag seems analogous to the

--target-dependents

flag for

pulumi up

. That flag allows the targeted deletion of a resource and anything that depends on that resource. I think that given the similarity in behavior there’s a pretty straightforward case to make here.