Data management API

Tier: Premium, Ultimate
Offering: GitLab Self-Managed
Status: Experiment

The availability of this feature is controlled by a feature flag. For more information, see the history.

Use the data management API to manage an instance’s data.

Prerequisites:

You must be an administrator.

Get information about a model

This endpoint is an experiment and might be changed or removed without notice.

GET /admin/data_management/:model_name

The :model_name parameter must be one of:

ci_job_artifacts
ci_pipeline_artifacts
ci_secure_files
container_repositories
dependency_proxy_blobs
dependency_proxy_manifests
design_management_repositories
group_wiki_repositories
lfs_objects
merge_request_diffs
packages_package_files
pages_deployments
projects
projects_wiki_repositories
snippet_repositories
terraform_state_versions
uploads

Supported attributes:

Attribute	Type	Required	Description
`model_name`	string	Yes	The pluralized name of the requested model. Must belong to the `:model_name` list above.
`checksum_state`	string	No	Search by checksum status. Allowed values: pending, started, succeeded, failed, disabled.
`identifiers`	array	No	Filter results with an array of unique identifiers of the requested model, which can be integers or base64 encoded strings.

This endpoint supports keyset pagination on the model’s primary key, with sorting in ascending or descending order. To use keyset pagination, add the pagination=keyset parameter to the request. By default, keyset pagination loads 20 records per page sorted in ascending order. You can modify the sorting order with the query parameter sort, and values asc or desc. To select a number of records per page, use the parameter per_page.

If successful, returns 200 and information about the model. It includes the following response attributes:

Attribute	Type	Description
`checksum_information`	JSON	Geo-specific checksum information, if available.
`created_at`	timestamp	Creation timestamp, if available.
`file_size`	integer	Size of the object, if available.
`model_class`	string	Class name of the model.
`record_identifier`	string or integer	Unique identifier of the record. Can be an integer or a base64 encoded string.

curl --header "PRIVATE-TOKEN: <your_access_token>" "https://primary.example.com/api/v4/admin/data_management/projects?pagination=keyset"

Example response:

[
  {
    "record_identifier": 1,
    "model_class": "Project",
    "created_at": "2025-02-05T11:27:10.173Z",
    "file_size": null,
    "checksum_information": {
      "checksum": "<object checksum>",
      "last_checksum": "2025-07-24T14:22:18.643Z",
      "checksum_state": "succeeded",
      "checksum_retry_count": 0,
      "checksum_retry_at": null,
      "checksum_failure": null
    }
  },
  {
    "record_identifier": 2,
    "model_class": "Project",
    "created_at": "2025-02-05T11:27:14.402Z",
    "file_size": null,
    "checksum_information": {
      "checksum": "<object checksum>",
      "last_checksum": "2025-07-24T14:22:18.214Z",
      "checksum_state": "succeeded",
      "checksum_retry_count": 0,
      "checksum_retry_at": null,
      "checksum_failure": null
    }
  }
]

Recalculate the checksum of selected model records

PUT /admin/data_management/:model_name/checksum

Attribute	Type	Required	Description
`model_name`	string	Yes	The pluralized name of the requested model. Must belong to the `:model_name` list above.
`checksum_state`	string	No	Filter by checksum status. Allowed values: pending, started, succeeded, failed, disabled.
`identifiers`	array	No	Filter records with an array of unique identifiers of the requested model, which can be integers or base64 encoded strings.

This endpoint marks all selected records from the model for checksum recalculation, filtered by checksum_state and identifiers parameters if provided. It enqueues a background job to do so. If successful, returns 200 and a JSON response containing the following information:

Attribute	Type	Description
`message`	string	A information message about the success or error.
`status`	string	Can be “success” or “error”.

curl --header "PRIVATE-TOKEN: <your_access_token>" "https://primary.example.com/api/v4/admin/data_management/projects/checksum"

Example response:

{
  "status": "success",
  "message": "Batch update job has been successfully enqueued."
}

Get information about a specific model record

GET /admin/data_management/:model_name/:id

Attribute	Type	Required	Description
`model_name`	string	Yes	The pluralized name of the requested model. Must belong to the `:model_name` list above.
`record_identifier`	string or integer	Yes	The unique identifier of the requested model. Can be an integer or a base64 encoded string.

If successful, returns 200 and information about the specific model record. It includes the following response attributes:

Attribute	Type	Description
`checksum_information`	JSON	Geo-specific checksum information, if available.
`created_at`	timestamp	Creation timestamp, if available.
`file_size`	integer	Size of the object, if available.
`model_class`	string	Class name of the model.
`record_identifier`	string or integer	Unique identifier of the record. Can be an integer or a base64 encoded string.

curl --header "PRIVATE-TOKEN: <your_access_token>" "https://primary.example.com/api/v4/admin/data_management/projects/1"

Example response:

{
  "record_identifier": 1,
  "model_class": "Project",
  "created_at": "2025-02-05T11:27:10.173Z",
  "file_size": null,
  "checksum_information": {
    "checksum": "<object checksum>",
    "last_checksum": "2025-07-24T14:22:18.643Z",
    "checksum_state": "succeeded",
    "checksum_retry_count": 0,
    "checksum_retry_at": null,
    "checksum_failure": null
  }
}

Recalculate the checksum of a specific model record

PUT /admin/data_management/:model_name/:record_identifier/checksum

Attribute	Type	Required	Description
`model_name`	string	Yes	The pluralized name of the requested model. Must belong to the `:model_name` list above.
`record_identifier`	string or integer	Yes	Unique identifier of the record. Can be an integer or a base64 encoded string (taken from the response of the GET query).

If successful, returns 200 and information about the specific model record. The checksum value is a representation of the queried model hashed with the md5 or sha256 algorithm.

curl --header "PRIVATE-TOKEN: <your_access_token>" "https://primary.example.com/api/v4/admin/data_management/projects/1/checksum"

Example response:

{
  "record_identifier": 1,
  "model_class": "Project",
  "created_at": "2025-02-05T11:27:10.173Z",
  "file_size": null,
  "checksum_information": {
    "checksum": "<sha256 or md5 string>",
    "last_checksum": "2025-07-24T14:22:18.643Z",
    "checksum_state": "succeeded",
    "checksum_retry_count": 0,
    "checksum_retry_at": null,
    "checksum_failure": null
  }
}