The issue with forcing `import_on_create` on the default worker pool is that if this resource has to be replaced (change of flavor for example), it will fail.

[vburckhardt]

          The issue with forcing `import_on_create` on the default worker pool is that if this resource has to be replaced (change of flavor for example), it will fail.

So maybe we could add a new field import_on_create to the var.worker_pools, then

import_on_create = each.value.pool_name == "default" ? each.value.import_on_create == null ? true : each.value.import_on_create : null

That way if the default workerpool needs replacement because of a flavor change, the user can set import_on_create = false and do the flavor change as well

Originally posted by @TwoDCube in #499 (comment)

[vburckhardt]

Opening this ticket to track creating an issue in the provider to get this limitation fixed

[ocofaigh]

Provider PR: IBM-Cloud/terraform-provider-ibm#5705

[Aashiq-J]

PR: #597

This PR adds support for orphan_on_delete which will help resolve the error of minimum 2 worker pool needed when deleting a cluster. But there is still a work around when it comes to re-creating a default worker pool, there is no way around it unless k8s team change their architecture.