Help Center/ Object Storage Service/ User Guide/ Data Security/ Using Versioning to Restore Data and Manage Object Versions/ Restoring Deleted Objects
Updated on 2026-09-23 GMT+08:00
View PDF
Share

Restoring Deleted Objects

After versioning is enabled, you can restore data lost due to accidental operations or application failures using the following methods:

  • Copy a specific historical object version and upload it to the bucket, making it the current version.
  • Permanently delete a delete marker so that the most recent historical version becomes the current version.

Prerequisites

  • Versioning is enabled for the bucket that stores the objects.
  • To recover data by permanently deleting a delete marker, ensure that the bucket contains at least one historical version of the same object without a delete marker.

Restoring a Deleted Object by Copying a Specific Historical Version

You can use APIs or SDKs to restore a deleted object through copying a specific historical version. You cannot use OBS Console, OBS Browser+, or obsutil to do so.

Before copying, obtain the required version ID by referring to Viewing Object Versions (When Versioning Is Enabled or Suspended).

Using APIs

Copying Objects

Using SDKs

Java

Python

C

Go

BrowserJS

.NET

Android

iOS

PHP

Node.js

Restoring a Deleted Object by Permanently Deleting a Delete Marker

Deleting an object from a versioning-enabled bucket does not remove the object itself. Instead, OBS inserts a delete marker to indicate that the latest version of the object is in the deleted state, while the previous current version becomes a historical version. You can restore a deleted object by permanently deleting the delete marker.

You can remove a delete marker using OBS Console, APIs, SDKs, or obsutil, but not OBS Browser+.

Using OBS Console

  1. In the navigation pane of OBS Console, choose Buckets.
  2. In the bucket list, click the desired bucket. The Objects page is displayed.
  3. Enable Historical Versions above the object list.
  4. In the Last Modified column, locate the object with Delete Marker and click Permanently Delete in the Operation column.

    You can also select all objects with Delete Marker and click Permanently Delete above the object list to recover them in a batch.

    Figure 1 Restoring a single object
    Figure 2 Batch restoring objects

  5. In the displayed dialog box, click OK.

    The delete marker disappears, and the Historical Version label next to the last modification time of the most recent historical version is removed. This indicates that the most recent historical version has become the current version, meaning the object has been recovered.

Using APIs

  1. List object versions in the bucket to obtain the version ID of the object that has a delete marker. In the response body, the object version that contains the DeleteMarker parameter is the delete marker, and its VersionId parameter indicates the version ID of that delete marker.
  2. Delete the object by specifying the version ID obtained in 1.

Using SDKs

  1. List object versions in the bucket to obtain the version ID of the object that has a delete marker. In the response body, the object version that contains the DeleteMarker parameter is the delete marker, and its VersionId parameter indicates the version ID of that delete marker. The SDKs for listing object versions are as follows.

    Java

    Python

    C

    Go

    BrowserJS

    .NET

    Android

    iOS

    PHP

    Node.js
  2. Delete the object by specifying the version ID obtained in 1. The SDKs for deleting an object are as follows.

    Java

    Python

    C

    Go

    BrowserJS

    .NET

    Android

    iOS

    PHP

    Node.js

Using the CLI Tool - obsutil

  1. List object versions in the bucket to obtain the version ID of the object that has a delete marker. In the response, VersionId under DeleteMarker list indicates the version ID of the delete marker.
  2. Delete the object by specifying the version ID obtained in 1. The command for deleting an object is as follows:

    Command Line Structure

    • Windows 
      obsutil rm obs://bucket/key [-f] -versionId=xxx [-fr] [-o=xxx] [-config=xxx] [-e=xxx] [-i=xxx] [-k=xxx] [-t=xxx]
    • macOS or Linux 
      ./obsutil rm obs://bucket/key [-f] -versionId=xxx [-fr] [-o=xxx] [-config=xxx] [-e=xxx] [-i=xxx] [-k=xxx] [-t=xxx]

    Examples

    In Windows, run obsutil rm obs://test-bucket/data_0001.txt -versionId=G001119ED32CD9F700004011024A1A22 to delete the delete marker of the data_0001.txt object in the bucket-test bucket. G001119ED32CD9F700004011024A1A22 is the version ID of the delete marker. 
    obsutil ls obs://test-bucket/data_0001.txt -v
Start at 2026-06-17 01:23:35.2387286 +0000 UTC

Listing versioning objects .

Versioning Object list:
Key                           VersionId                               LastModified                  Size      StorageClass        ETag
obs://test-bucket/data_0001.txt
                              G001119ED0A0BF6800004017005E42D5        2026-06-16T13:30:56Z          0B        standard            "d41d8cd98f00b204e9800998ecf8427e"

DeleteMarker list:
Key                           VersionId                               LastModified                  StorageClass
obs://test-bucket/data_0001.txt
                              G001119ED32CD9F700004011024A1A22        2026-06-17T01:23:12Z          standard

Total size of prefix [data_0001.txt]: 0B
Folder number: 0
Versioning file number: 1
DeleteMarker number: 1

obsutil rm obs://test-bucket/data_0001.txt -versionId=G001119ED32CD9F700004011024A1A22
Start at 2026-06-17 01:23:56.7597831 +0000 UTC

Do you want delete object [data_0001.txt] in the bucket [test-bucket]? Please input (y/n) to confirm:
y
Delete object [data_0001.txt] with version id [G001119ED32CD9F700004011024A1A22] in the bucket [test-bucket] successfully, cost [187], request id [0000019ED32D911E4015141E86F59409]

obsutil ls obs://test-bucket/data_0001.txt -v
Start at 2026-06-17 01:24:03.869759 +0000 UTC

Listing versioning objects .

Versioning Object list:
Key                           VersionId                               LastModified                  Size      StorageClass        ETag
obs://test-bucket/data_0001.txt
                              G001119ED0A0BF6800004017005E42D5        2026-06-16T13:30:56Z          0B        standard            "d41d8cd98f00b204e9800998ecf8427e"

Total size of prefix [data_0001.txt]: 0B
Folder number: 0
Versioning file number: 1
DeleteMarker number: 0

    Parameter Description

    Parameter

    Optional or Mandatory

    Description

    bucket

    Mandatory

    The bucket name

    key

    Mandatory for deleting a single object.

    Optional for deleting objects in batches.

    The name of the object to be deleted, or the name prefix of the objects to be deleted in batches

    NOTE:

    If this parameter is left blank when deleting objects in batches, all objects in the bucket are deleted. Use this parameter with caution.

    fr

    Optional for deleting a single object (additional parameter)

    Generates an operation result file when deleting an object.

    f

    Optional (additional parameter)

    Performs operations without asking for confirmation. Use this parameter with caution.

    versionId

    Optional for deleting a single object (additional parameter)

    Mandatory for deleting object versions

    The version ID of the object to be deleted.

    r

    Mandatory for deleting objects in batches (additional parameter)

    Deletes objects in batches based on a specified object name prefix.

    CAUTION:

    When you batch delete objects, all objects with the specified prefix will be deleted.

    j

    Optional for deleting objects in batches (additional parameter)

    The maximum number of concurrent tasks for deleting objects in batches. The default value is the value of defaultJobs in the configuration file.

    NOTE:

    The tool ensures that this value is at least 1.

    v

    Optional for deleting objects in batches (additional parameter)

    Deletes versions of an object and the delete markers in batches based on a specified object name prefix.

    o

    Optional (additional parameter)

    The folder that stores the result files. After the command is executed, result files (possibly success and failure files) will be created in the specified folder. The default value is .obsutil_output, a subfolder in the user's home directory where obsutil commands are executed.

    NOTE:
    • A result file should be named as follows: rm_{succeed | failed}_report_Time_TaskId.txt.
    • By default, the maximum size of a single result file is 30 MB and up to 1,024 result files can be retained. You can modify these limits by configuring recordMaxLogSize and recordBackups in the configuration file.
    • If there are multiple folders and files and you need to confirm the details of a failed task, refer to the failure result file cp_failed_report_Time_TaskId.txt in the result folder and the log files in the log path.

    bucket-cname

    Optional (additional parameter)

    The user-defined domain name bound to the bucket

    NOTE:

    This parameter is only supported by obsutil 5.7.9 and later.

    config

    Optional (additional parameter)

    The user-defined configuration file for executing the current command. For details about parameters that can be configured, see Configuration Parameters.

    e

    Optional (additional parameter)

    The endpoint

    i

    Optional (additional parameter)

    The user's AK

    k

    Optional (additional parameter)

    The user's SK

    t

    Optional (additional parameter)

    The user's security token

    Response

    Field

    Description

    Parallel

    The parameter -p in the request

    Jobs

    The parameter -j in the request

    Threshold

    The parameter -threshold in the request

    PartSize

    The parameter -ps in the request

    Exclude

    The parameter -exclude in the request

    Include

    The parameter -include in the request

    TimeRange

    The parameter -timeRange in the request

    VerifyLength

    The parameter -vlength in the request

    VerifyMd5

    The parameter -vmd5 in the request

    CheckpointDir

    The parameter -cpd in the request

    OutputDir

    The parameter -o in the request

    ArcDir

    The parameter -arcDir in the request

    Succeed count

    The number of successful tasks

    Failed count

    The number of failed tasks

    Skip count

    The number of tasks that are skipped during incremental upload, download, or copy, and synchronous upload, download, or copy.

    NOTE:

    Skipped tasks are recorded into successful tasks.

    Warning count

    The number of tasks that are executed successfully but contain warnings.

    NOTE:
    • The task for which a warning is generated may be a failure or a success, which needs to be further determined according to the corresponding result list.
    • The number of tasks that generate warnings is independent of the number of successful or failed tasks. The total number of tasks is the number of successful tasks plus the number of failed tasks.

    Succeed bytes

    The number of bytes that are successfully uploaded

    max cost

    The maximum duration of all tasks, in ms

    min cost

    The minimum duration of all tasks, in ms

    average cost

    The average duration of all tasks, in ms

    average tps

    The average number of tasks completed per second

    Task id

    The unique ID of an operation, which is used to search for the result file generated for a batch task
Parent Topic: Using Versioning to Restore Data and Manage Object Versions