Skip to content

2026-11-19

Reworked S3 Service Deactivation

Remaining buckets in an S3 Service Instance no longer block the deactivation of the S3 Service in the Portal and through its API. Previously, you had to delete all buckets and objects thereinmanually before you could deactivate the Service.

Now the deactivation permanently deletes the remaining buckets and their objects, including buckets that contain locked objects due to an object locking policy. Deleted buckets and objects cannot be recovered. Billing for the S3 Service stops at the deactivation. The deactivation dialog in the Project Settings now states that all existing buckets and their objects will be deleted.

Deactivate an S3 Service via the API

The deactivate endpoint for S3 Services now also accepts requests while buckets remain. If force is missing or false, the API returns 409 Conflict and does not deactivate the Service, because the deactivation would delete the remaining buckets. Set force=true to confirm and start the deactivation.

Enhancements and Bug Fixes

  • Enhancement - More reliable deactivation of Compute Services: We have adjusted how the backend handles the deactivation of Compute Services, so that deactivations are processed reliably.
  • Bug Fix - Missing transitional states in resource lists: Fixed an issue where S3 access keys, S3 buckets, and the members of Compute and Kubernetes Service Instances did not show their transitional state, such as "Creating" or "Deleting", while a change was in progress.
  • Bug Fix - Delayed confirmation when creating Shared Application Credentials: Fixed an issue where the confirmation message appeared only after the Portal had finished generating the credential.
  • Bug Fix - Server error when creating a Shared Application Credential with a blank name: Fixed an issue where entering only whitespace as the name of a Shared Application Credential caused a server error.
  • Bug Fix - Compute Service Instance member stuck in "Creating": Fixed an issue where a Compute Service Instance member could remain in the "Creating" state indefinitely. This happened when the user already had an OpenStack account and was already a member of the corresponding OpenStack project, but the Portal had not yet linked the two accounts.