sedefsavas

We have a new e2e test for unmanaged infra, we can add a check there to confirm cloud provider tags we add are gone after the cluster is deleted right...

@pydctw What is the status of this PR?

We can add that permission to CloudFormation policies in `clusterawsadm`.

Since this is a new support, we could document the requirement to rerun clusterawsadm or add the permission manually to the IAMs. cc @richardcase

You will need to sign the CLA by clicking on the failed CLA job. If this is ready for another review, you can delete the WIP from the title.

Failing to reconcile when additional SGs could not be found is a change of behaviour, so we can do this change in the v1beta2 version. It'd be good to track...

Based on[ this comment ](https://github.com/kubernetes-sigs/cluster-api-provider-aws/pull/3526#issuecomment-1154556309)above, let's wait until we open the main branch to v1beta2 changes before rebasing to avoid double efforts.

/lifecycle frozen

No, this is not related to SDK calls. We cache providers/sessions but do not clean them. For example, identities that are no longer in use will linger until CAPA controller...

/ok-to-test