Change storage class and restore archives

The storage class determines how the service stores an object’s data and makes it accessible. An archive class may require restoration before reading, downloading or copying. Restoration makes content accessible; it is not a storage class change.

A class change performed by rewriting follows S3 copy conditions. If the applicable source condition is disabled, a concurrent change after preliminary verification may be replaced. An archive restore without copying remains independent.

Check capabilities and state#

Open File > Accounts…, select the account, click Edit…, then open Capabilities > Storage classes. This configuration determines the available classes, permitted write operations and restore options: durations and retrieval tiers. Account default policy and Assignments determine the location’s policy. Configure only features the service actually supports: this creates neither storage classes nor remote permissions.

The Amazon S3 general purpose bucket policy includes STANDARD, REDUCED_REDUNDANCY, STANDARD_IA, ONEZONE_IA, INTELLIGENT_TIERING, GLACIER_IR, GLACIER and DEEP_ARCHIVE. Compatible services may offer other values.

1. In the S3 pane, select the object or its specific version with Show versions.
2. Open Properties > Storage.
3. Read Current storage class: and Archived content access:.

Storage class not configured for this account requires correcting the local policy before an operation that depends on this class. The browser’s storage class column may also show an archive suffix; properties provide more precise information.

When the response and lifecycle policy allow it, Lifecycle expiration date: and Lifecycle rule ID: show the date and rule, with a button to copy the rule identifier. Not reported by the service does not prove that no rule exists; Invalid expiration information returned by the service indicates an unusable value. See lifecycle.

If a listing omits the class, S3Client keeps it unknown and reads metadata through HeadObject before a download, preview, restore or rewrite that depends on it. The metadata read must succeed; an unconfigured class, conflicting class values or unusable archive state blocks the action. If HEAD also omits the class, S3Client uses Default class from the applicable policy.

Change the current storage class#

Changing the class rewrites the object through a copy. It may create a new version, change the date and incur request, retrieval or storage charges. Earlier versions retain their class and still occupy storage.

The source must be readable. Usual permissions are s3:GetObject and s3:PutObject, plus permissions for encryption, tags, ACLs and protections to preserve. Large objects may require multipart copying.

If Enable ACL operations for this account is unchecked, changing the storage class requires no ACL method or permission: the copy uses the destination default without ACL requests or headers. Other permissions are still required. Preservation of the existing ACL is not guaranteed if the service still uses it, even within the same bucket. See local ACL deactivation.

1. Select the current object with Show versions off, or the latest version with it on. In the latter case, check that General > Latest version: shows Yes (from the last version listing).
2. Open Properties > Storage.
3. Choose the destination class in New storage class:.
4. Check other pending changes and preservation warnings.
5. Click OK to start the copy.
6. After completion, refresh the pane and open Storage on the current object or the new version; check Current storage class:.

(default) lets the service apply its default class as described by the local policy. An observed class that is not offered as a destination for this change appears with (current). Choices require Change storage class on the same object for the destination class, together with a permitted copy path in the account policy. S3Client checks the other transfer requirements during execution.

If no different destination class is available, New storage class: is disabled with Changing the storage class on the same object is not enabled for an available destination class. Check service support before changing Account > Capabilities > Storage classes. Both Google presets disable this permission: you can choose a class for an upload or a copy to another key, but cannot change the class on the same key in S3Client.

In version view, New storage class: is available for the latest version when the other requirements are met. Before copying, S3Client checks that the selected version is still current; if it has been replaced, the operation stops. The copy writes to the same key. A source version with an immutable identifier keeps its class; with versioning suspended, the null version may be replaced. For an older version, New storage class: remains disabled: copy it to the current object or another destination to apply new settings.

Restore an archive#

Restoration normally requires s3:RestoreObject and support for this class and target. It can incur charges and take time. A successful request does not mean the content is already readable.

If the archive state has not yet been observed, S3Client first reads the metadata with HeadObject. This normally requires s3:GetObject, or s3:GetObjectVersion for a specific version, and the SSE-C key if the object uses that encryption. HeadObject permissions.

1. Select the archived object or its specific version.
2. Use Restore archived object… in the context menu when available. Alternatively, open Properties > Storage and select Restore archived object.
3. Check Target: Current object or Version followed by its identifier.
4. Enter the available parameters.
5. Confirm the dialog or properties with OK.
6. After the request, refresh and check Archived content access: until the content becomes readable.

See AWS retrieval options.

For Amazon S3 GLACIER and DEEP_ARCHIVE, a temporary copy becomes readable without changing class; duration is required. DEEP_ARCHIVE does not offer Expedited. For INTELLIGENT_TIERING archive tiers, content returns to a directly accessible tier without a temporary duration. GLACIER_IR is directly readable in the Amazon S3 policy.

The INTELLIGENT_TIERING Deep Archive Access tier also does not support Expedited. AWS retrieval options.

The requested duration does not protect the object from lifecycle expiration: on Amazon S3, an earlier expiration can delete the object before that duration ends. Amazon S3 restoration.

A request already in progress cannot be repeated from properties. A specific version requires support in the policy. An unrecognized class, unusable state or missing source can block the request; address the reported issue and refresh.

Submit restoration and tag changes separately. Also wait for readable content before rewriting metadata or storage class. To apply a class to current objects in bulk, see child objects.

For several selected objects, open Properties and the relevant tab. Observed values may differ; only your explicit changes are applied. See properties for multiple objects.