# Object Deletion

LLMS index: [llms.txt](/llms.txt)

---

<a id="object-deletion"></a>
<a id="minio-object-delete"></a>

## Overview {#overview}

This page summarizes how a `DELETE` operation affects objects depending on the configuration of the bucket that contains the object.

Any combination of the following factors may impact how `DELETE` operations function:

- [Bucket versioning](/administration/object-management/object-versioning/#minio-bucket-versioning)
- [Object locking rules](/administration/object-management/object-retention/#minio-object-locking)
- [Object Lifecycle Management rules](/administration/object-management/object-lifecycle-management/#minio-lifecycle-management)
- [Object tiering](/administration/object-management/object-lifecycle-management/#minio-lifecycle-management-tiering)
- [Site](/operations/replication/multi-site-replication/#minio-site-replication-overview) or [bucket](/administration/bucket-replication/#minio-replication-behavior-delete) replication
- [Scanner](/operations/concepts/scanner/#minio-concepts-scanner)

## Permissions {#permissions}

MinIO uses a [policy based access control](/administration/identity-access-management/policy-based-access-control/#minio-policy) system for access management. The user or service account must provide the correct policy action and conditions to allow a `DELETE` for the bucket and object.

The required action is selected from the request, not merely from whether the
bucket has versioning enabled:

| Request                        | Required action                            |
| ------------------------------ | ------------------------------------------ |
| `DELETE` without `versionId`   | `s3:DeleteObject`                          |
| `DELETE` with a version UUID   | `s3:DeleteObjectVersion`                   |
| `DELETE` with `versionId=null` | `s3:DeleteObjectVersion`                   |
| `DeleteObjects`                | Evaluated independently for each XML entry |
{.wide .expand}

An explicit deny always takes precedence. Conditions using `s3:versionid` see
the effective version ID of the object being deleted; a query parameter on the
outer `DeleteObjects` request does not replace an entry's `VersionId`.

## Unversioned Objects {#unversioned-objects}

When performing a `DELETE` operation on an object in a bucket that does not have versioning enabled, the operation is straightforward. After verifying the user or service account has permission to perform the `DELETE` operation, MinIO permanently removes the object.

The user or service account requesting the delete action must have the [`s3:DeleteObject`](/administration/identity-access-management/policy-based-access-control/#policy-action.s3-DeleteObject) action permission for the bucket and object.

## Versioned Objects {#versioned-objects}

`DELETE` operations work differently when an object is versioned.

Deleting a specifically named version requires
[`s3:DeleteObjectVersion`](/administration/identity-access-management/policy-based-access-control/#policy-action.s3-DeleteObjectVersion).
Deleting without a version ID still requires `s3:DeleteObject` and creates a
delete marker instead of removing an existing version.

### Delete operations on the current version {#delete-operations-on-the-current-version}

A `DELETE` operation on a versioned object that does not specify a version UUID results in the creation of a `DeleteMarker` placed as the `head` of the object.

In this scenario, MinIO does not actually remove the object or any of its versions from the disk. All existing versions of the object remain available to access by specifying the version’s UUID. When a `DeleteMarker` is the head for the object, MinIO does not serve the object for `GET` requests that do not specify a version ID. Instead, MinIO returns a `404`-like response.

You can find the UUID of object versions with [`mc ls --versions`](/reference/minio-mc/mc-ls/#mc.ls.-versions).

To remove the current version of the object from the drive, find the UUID of the version, and then use [`mc rm --version-id=UUID ...`](/reference/minio-mc/mc-rm/#mc.rm.-version-id) to delete the current version. In this scenario, the immediately preceding version of the object then becomes the current version of the object served for `GET` requests of the object with no UUID specified.

> [!CAUTION]
> **Warning**
>
> Specifying a `version-id` in a DELETE operation is irreversible. MinIO removes the specified version from the drive and **cannot** retrieve it.

### Delete operations on a prior version {#delete-operations-on-a-prior-version}

To delete prior versions of an object, specify the version’s UUID. You can retrieve the version UUID with [`mc ls --versions`](/reference/minio-mc/mc-ls/#mc.ls.-versions). When the `DELETE` request specifies a `version-id` and the user has the correct permissions to delete the object version, MinIO permanently removes the specified version from the drive.

> [!CAUTION]
> **Warning**
>
> Specifying a `version-id` in a DELETE operation is irreversible. MinIO removes the specified version from the drive and **cannot** retrieve it.

### Delete all versions {#delete-all-versions}

Use [`mc rm --versions`](/reference/minio-mc/mc-rm/#mc.rm.-versions) to delete *all* versions of an object. This is irreversible.

## Lifecycle Management Expiration {#lifecycle-management-expiration}

You can define one or more [lifecycle management expiration rule(s)](/administration/object-management/create-lifecycle-management-expiration-rule/#minio-lifecycle-management-create-expiry-rule) to expire objects after a certain version number count or a certain period of time. When more versions exist than the rule specifies, or when a version is older than specified, MinIO permanently removes the object version from the drive.

These rules rely on the [scanner](/operations/concepts/scanner/#minio-concepts-scanner) to process the rule on the bucket. The scanner operates as a lower priority continuous process where `READ` and `WRITE` actions are preferred. Because of this, object versions that meet the requirements for expiration may not immediately be removed from MinIO.

See the [scanner](/operations/concepts/scanner/#minio-concepts-scanner) page for more details on how the scanner works and configuration options.

`DeleteMarkers` are their own objects. Lifecycle rules can remove `DeleteMarkers` that are the only remaining versions of their objects.

> [!NOTE]
> **Changed: MinIO**
>
> RELEASE.2024-05-01T01-11-10Z

With `JSON`, lifecycle rules can remove all versions of a deleted object after a specified number of days.

Starting in [SILO 20261109](/blog/release/silo-20261109/), lifecycle expiration enforces the following safeguards:

- **Exact-key delete-all scope:** `ExpiredObjectAllVersions` and `DelMarkerExpiration` act only on the expired object key rather than child keys below that prefix, and refuse deletion when the key still holds tiered or Object Lock protected versions ([#297](https://github.com/pgsty/silo/issues/297)).
- **`DelMarkerExpiration`-only rules:** Rules whose sole action is `DelMarkerExpiration` execute as configured ([#268](https://github.com/pgsty/silo/issues/268)).
- **Multi-pool quorum and restored-copy expiry:** Lifecycle expiry evaluates under the object lock across all server pools and requires an authoritative quorum before deleting ([#278](https://github.com/pgsty/silo/issues/278)). Expiring a restored copy removes only the temporary local cache while keeping the object version and its remote-tier data intact ([#281](https://github.com/pgsty/silo/issues/281)).

## Retained Objects {#retained-objects}

MinIO protects objects subject to a [locking rule](/administration/object-management/object-retention/#minio-object-locking) from being overwritten or deleted. These rules require that objects be retained until either the rule expires or is removed.

`DELETE` operations on locked objects without a specified version result in the creation of a *DeleteMarker* for the object. However, the object versions themselves are retained as required by the lock.

`DELETE` operations that specify an object version are subject to the retention rules. MinIO protects object versions subject to a lock from being overwritten or deleted until the lock expires or is removed.

In [SILO 20261109](/blog/release/silo-20261109/), bulk `DeleteObjects` requests check physical version removals sequentially under the same mutation lock used for retention and legal-hold updates, merging protection across all server pools (commit [`6ef3a52fc`](https://github.com/pgsty/silo/commit/6ef3a52fc); large version-purge batches in Object Lock buckets therefore take longer, while delete-marker creation and ordinary buckets retain the concurrent bulk path). Bulk deletion also checks Object Lock retention and legal hold on directory objects before removing a directory `null` version (commit [`aa2d79daf`](https://github.com/pgsty/silo/commit/aa2d79daf)).

## Replicated Objects {#replicated-objects}

Replication duplicates objects from one location to another. MinIO supports replication at the bucket level or the cluster (“site”) level.

Delete operations may or may not replicate, depending on the type of replication and how the replication is configured.

In [SILO 20261109](/blog/release/silo-20261109/), replicated deletes preserve the target's legal hold and retention (commit [`7e9ee6bef`](https://github.com/pgsty/silo/commit/7e9ee6bef)): a source purge that bypassed `GOVERNANCE` retention does not bypass active protection on the target copy, so the target refuses the purge and the source retries. In addition, lifecycle-created delete markers and version purges are **not** replicated across bucket or site replication (commit [`1bd0b1b04`](https://github.com/pgsty/silo/commit/1bd0b1b04)), matching AWS S3 semantics where each replica bucket applies its own lifecycle configuration independently.

### Site Replication {#site-replication}

For clusters with [multi-site replication](/operations/replication/multi-site-replication/#minio-site-replication-overview) enabled, SILO replicates explicit client `DELETE` operations performed on any cluster to each peer site in the group. Lifecycle-created delete markers and version purges do not replicate across sites (commit [`1bd0b1b04`](https://github.com/pgsty/silo/commit/1bd0b1b04)).

Delete behavior on any single peer follows the same processes as any SILO deployment.

### Bucket Replication {#bucket-replication}

With [bucket replication](/administration/bucket-replication/#minio-bucket-replication), MinIO supports replicating delete operations between a source bucket and a configured remote bucket. MinIO synchronizes deleting specific object versions *and* new [delete markers](https://docs.aws.amazon.com/AmazonS3/latest/userguide/delete-marker-replication.html). Delete operation replication uses the same [replication process](/administration/bucket-replication/#minio-replication-process) as all other replication operations.

MinIO requires *explicitly enabling* versioned deletes and delete marker replication. Use the [`mc replicate add --replicate`](/reference/minio-mc/mc-replicate-add/#mc.replicate.add.-replicate) field to specify either `delete` and `delete-marker` or both to enable versioned deletes and delete marker replication, respectively. To enable both, specify both strings using a comma separator: `delete,delete-marker`.

For delete marker replication, MinIO begins the replication process after a delete operation creates the delete marker. MinIO uses the `X-Minio-Replication-DeleteMarker-Status` metadata field for tracking delete marker replication status. In [active-active](/administration/bucket-replication/enable-server-side-two-way-bucket-replication/#minio-bucket-replication-serverside-twoway) replication configurations, MinIO may produce duplicate delete markers if both clusters concurrently create a delete marker for an object *or* if one or both clusters were down before the replication event synchronized.

For replicating the deletion of a specific object version, MinIO marks the object version as `PENDING` until replication completes. Once the remote target deletes that object version, MinIO deletes the object version on the source. While this process ensures near-synchronized version deletion, it may result in listing operations returning the object version after the initial delete operation. MinIO uses the `X-Minio-Replication-Delete-Status` for tracking delete version replication status.

MinIO only replicates explicit client-driven delete operations. MinIO does *not* replicate objects deleted by [lifecycle management expiration rules](/administration/object-management/object-lifecycle-management/#minio-lifecycle-management-expiration). For [active-active](/administration/bucket-replication/enable-server-side-two-way-bucket-replication/#minio-bucket-replication-serverside-twoway) configurations, set the same expiration rules on *all* of the replication buckets to ensure consistent application of object expiration.

> [!DETAILS]- MinIO Trims Empty Object Prefixes on Source and Remote Bucket
> If a delete operation removes the last object in a bucket prefix, MinIO recursively removes each empty part of the prefix up to the bucket root. MinIO only applies the recursive removal to prefixes created *implicitly* as part of object write operations. MinIO does not recursively remove prefixes created using an explicit directory creation command, such as [`mc mb`](/reference/minio-mc/mc-mb/#command-mc.mb).
>
> If a replication rule enables replication delete operations, the replication process *also* applies the implicit prefix trimming behavior on the destination MinIO cluster.
>
> For example, consider a bucket `photos` with the following object prefixes:
>
> - `photos/2021/january/myphoto.jpg` // `2021/january/` created implicitly based on the object name
> - `photos/2021/february/myotherphoto.jpg` // `2021/february/` created implicitly based on the object name
> - `photos/NYE21/NewYears.jpg` // `NYE21/` explicitly created in the bucket
>
> `photos/NYE21` is the *only* prefix explicitly created using [`mc mb`](/reference/minio-mc/mc-mb/#command-mc.mb). All other prefixes were *implicitly* created as part of writing the object located at that prefix.
>
> - A command removes `myphoto.jpg`. MinIO automatically trims the empty `/january/` prefix.
> - A command then removes the `myotherphoto.jpg`. MinIO automatically trims the `/february/` prefix *and* the now-empty `/2021` prefix.
> - A command removes the `NewYears.jpg` object. MinIO leaves the `/NYE21/` prefix in place since it was *explicitly* created.