Delete a bucket

Delete a bucket to remove it and all its objects for good. It also deletes the bucket’s access key and ends its storage charge. The portal’s messages call the bucket an Object Storage dataset, or a dataset.

Before you begin

• Permission to delete datasets in the project. A Member gets it through a policy that allows deleting datasets.
• Copies of any objects you need. Deleting is permanent. To keep the data on a Block Storage disk, create a disk and copy the data across.
• The bucket unmounted from every workload. The portal refuses the delete while the bucket is mounted on an instance or Virtual Machine. It also refuses while a fine-tuning job that has not succeeded, a model endpoint, a deployment, or a Slurm Cluster uses the bucket. This holds whether the workload is running or stopped. The error names what holds the bucket.
• Any replication rule deleted; see Replicate a bucket to another region.
• Any default object lock removed; see Set a default object lock on a bucket.
• A different name in mind if you want to create a bucket again. If the form says the name is taken, choose a different one.

Delete the bucket and its Object Storage dataset

1. In the list, open the Actions menu of the bucket and select Delete. For a bucket TIR created, the delete permanently removes the bucket and all its objects.
2. Tick I understand and want to delete <alias>.
3. Select Delete. The portal shows “Object storage <alias> deleted successfully”.

If an existing EOS bucket was attached to TIR, the delete removes it from TIR and leaves that bucket itself in place, and the dialog says so. The dialog also says the bucket stays if another dataset is using it.

Check the result

The bucket leaves the list. Buckets mounted on an instance stay when you delete the instance. A bucket keeps its storage charge until you delete it here.

For a bucket you only want to clear, empty it instead. For help with a refused delete, see Troubleshooting TIR Object Storage.