Elasticsearch privileges

Stack Serverless

This section provides detailed reference information for Elasticsearch privileges.

If you're using a stack-versioned deployment such as a self-managed cluster, Elastic Cloud Hosted, Elastic Cloud on Kubernetes, or Elastic Cloud Enterprise, then refer to User roles for more information on how role-based access control works.

If you're using Elastic Cloud Serverless, then refer to Serverless project custom roles to learn how to build roles using these privileges.

Roles are governed by a set of configurable privileges grouped into these categories:

• cluster, which you can use to manage core operations like snapshots, managing API keys, autoscaling, and cross-cluster functionality.
• indices, which govern document-level access, index and data stream metadata information, and more.
• run-as, which allows for secure impersonation.
• application, which enable external applications to define and store their privilege models within Elasticsearch roles.

When creating roles, refer to this page for a complete list of available privileges.

all

All cluster administration operations, like snapshotting, node shutdown/restart, settings update, rerouting, or managing users and roles.

cancel_task

Privileges to cancel tasks and delete async searches. See delete async search API for more informations.

create_snapshot Serverless Unavailable

Privileges to create snapshots for existing repositories. Can also list and view details on existing repositories and snapshots.

cross_cluster_replication Serverless Unavailable

Privileges to connect to remote clusters configured with the API key based model for cross-cluster replication.

Note

This privilege must not be directly granted. It is used internally by Create Cross-Cluster API key and Update Cross-Cluster API key to manage cross-cluster API keys.

cross_cluster_search Serverless Unavailable

Privileges to connect to remote clusters configured with the API key based model for cross-cluster search.

Note

This privilege must not be directly granted. It is used internally by Create Cross-Cluster API key and Update Cross-Cluster API key to manage cross-cluster API keys.

grant_api_key Serverless Unavailable

Privileges to create Elasticsearch API keys on behalf of other users.

manage

Builds on monitor and adds cluster operations that change values in the cluster. This includes snapshotting, updating settings, and rerouting. It also includes obtaining snapshot and restore status. This privilege does not include the ability to manage security.

manage_api_key

All security-related operations on Elasticsearch REST API keys including creating new API keys, retrieving information about API keys, querying API keys, updating API key, bulk updating API keys, and invalidating API keys.

Note

• When you create new API keys, they will always be owned by the authenticated user.
• When you have this privilege, you can invalidate your own API keys and those owned by other users.

manage_autoscaling Serverless Unavailable

All operations related to managing autoscaling policies.

manage_ccr Serverless Unavailable

All cross-cluster replication operations related to managing follower indices and auto-follow patterns. It also includes the authority to grant the privileges necessary to manage follower indices and auto-follow patterns. This privilege is necessary only on clusters that contain follower indices.

manage_data_frame_transforms Serverless Unavailable Stack Deprecated 7.5.0

Use manage_transform instead. Elasticsearch version 7.5 and older used manage_data_frame_transforms in operations related to managing transforms.

manage_data_stream_global_retention Stack Deprecated 8.16.0

This privilege has no effect.

manage_enrich

All operations related to managing and executing enrich policies.

manage_ilm Serverless Unavailable

All index lifecycle management operations related to managing policies.

manage_index_templates

All operations on index templates.

manage_inference

All operations related to managing inference.

manage_ingest_pipelines

All operations on ingest pipelines.

manage_logstash_pipelines

All operations on logstash pipelines.

manage_ml

All machine learning operations, such as creating and deleting datafeeds, jobs, and model snapshots.

Note

Datafeeds that were created prior to version 6.2 or created when security features were disabled run as a system user with elevated privileges, including permission to read all indices. Newer datafeeds run with the security roles of the user who created or updated them.

manage_oidc Serverless Unavailable

Enables the use of Elasticsearch APIs (OpenID connect prepare authentication, OpenID connect authenticate, and OpenID connect logout) to initiate and manage OpenID Connect authentication on behalf of other users.

manage_own_api_key

All security-related operations on Elasticsearch API keys that are owned by the current authenticated user. The operations include creating new API keys, retrieving information about API keys, querying API keys, updating API key, bulk updating API keys, and invalidating API keys.

manage_pipeline

All operations on ingest pipelines.

manage_rollup Serverless Unavailable

All rollup operations, including creating, starting, stopping and deleting rollup jobs.

manage_saml Serverless Unavailable

Enables the use of internal Elasticsearch APIs to initiate and manage SAML authentication on behalf of other users.

manage_search_application

All CRUD operations on search applications.

manage_search_query_rules

All CRUD operations on query rules.

manage_search_synonyms

All synonyms management operations on Synonyms APIs.

manage_security

All security-related operations such as CRUD operations on users and roles and cache clearing.

manage_service_account Serverless Unavailable

All security-related operations on Elasticsearch service accounts including Get service accounts, Create service account tokens, Delete service account token, and Get service account credentials.

manage_slm Serverless Unavailable Stack Deprecated 8.15.0

All snapshot lifecycle management (SLM) actions, including creating and updating policies and starting and stopping SLM. It also grants the permission to start and stop index lifecycle management, using the ILM start and ILM stop APIs.

In a future major release, this privilege will not grant any index lifecycle management permissions.

manage_token Serverless Unavailable

All security-related operations on tokens that are generated by the Elasticsearch Token Service.

manage_transform

All operations related to managing transforms.

manage_watcher Serverless Unavailable

All watcher operations, such as putting watches, executing, activate or acknowledging.

Note

Watches that were created prior to version 6.1 or created when the security features were disabled run as a system user with elevated privileges, including permission to read and write all indices. Newer watches run with the security roles of the user who created or updated them.

monitor

All cluster read-only operations, like cluster health and state, hot threads, node info, node and cluster stats, and pending cluster tasks.

monitor_data_stream_global_retention Stack Deprecated 8.16.0 Stack Unavailable

This privilege has no effect.

monitor_enrich

All read-only operations related to managing and executing enrich policies.

monitor_esql Stack 9.1.0

All read-only operations related to ES|QL queries.

monitor_inference

All read-only operations related to inference.

monitor_ml

All read-only machine learning operations, such as getting information about datafeeds, jobs, model snapshots, or results.

monitor_rollup Serverless Unavailable

All read-only rollup operations, such as viewing the list of historical and currently running rollup jobs and their capabilities.

monitor_snapshot Serverless Unavailable

Privileges to list and view details on existing repositories and snapshots.

monitor_stats Serverless Unavailable

Privileges to list and view details of stats.

monitor_text_structure Serverless Unavailable

All read-only operations related to the find structure API.

monitor_transform

All read-only operations related to transforms.

monitor_watcher Serverless Unavailable

All read-only watcher operations, such as getting a watch and watcher stats.

read_ccr Serverless Unavailable

All read-only cross-cluster replication operations, such as getting information about indices and metadata for leader indices in the cluster. It also includes the authority to check whether users have the appropriate privileges to follow leader indices. This privilege is necessary only on clusters that contain leader indices.

read_ilm Serverless Unavailable

All read-only index lifecycle management operations, such as getting policies and checking the status of {Ilm}

read_pipeline

Read-only access to ingest pipeline (get, simulate).

read_slm Serverless Unavailable Stack Deprecated 8.15.0

All read-only SLM actions, such as getting policies and checking the SLM status. It also grants the permission to get the index lifecycle management status, using the ILM get status API.

In a future major release, this privilege will not grant any index lifecycle management permissions.

read_security

All read-only security-related operations, such as getting users, user profiles, Elasticsearch API keys, Elasticsearch service accounts, roles and role mappings. Allows querying and retrieving information on all Elasticsearch API keys.

transport_client Serverless Unavailable

All privileges necessary for a transport client to connect. Required by the remote cluster to enable cross-cluster search.

all

Any action on an index or data stream.

auto_configure

Permits auto-creation of indices and data streams. An auto-create action is the result of an index or bulk request that targets a non-existent index or data stream rather than an explicit create index or create data stream request. Also permits auto-update of mappings on indices and data streams if they do not contradict existing mappings. An auto-update mapping action is the result of an index or bulk request on an index or data stream that contains new fields that may be mapped rather than an explicit update mapping request.

create

Privilege to index documents.

Note

This privilege does not restrict the index operation to the creation of documents but instead restricts API use to the index API. The index API allows a user to overwrite a previously indexed document. See the create_doc privilege for an alternative.

Important

Starting from 8.0, this privilege no longer grants the permission to update index mappings. In earlier versions, it implicitly permitted index mapping updates (excluding data stream mappings) via the updating mapping API or through dynamic field mapping. Mapping update capabilities will be fully removed in a future major release.

create_doc

Privilege to index documents. It does not grant the permission to update or overwrite existing documents.

Note

This privilege relies on the op_type of indexing requests (Index and Bulk). When ingesting documents as a user who has the create_doc privilege (and no higher privilege such as index or write), you must ensure that op_type is set to create through one of the following:

• Explicitly setting the op_type in the index or bulk APIs
• Using the _create endpoint for the index API
• Creating a document with an auto-generated _id

Important

Starting from 8.0, this privilege no longer grants the permission to update index mappings. In earlier versions, it implicitly permitted index mapping updates (excluding data stream mappings) via the updating mapping API or through dynamic field mapping. Mapping update capabilities will be fully removed in a future major release.

create_index Serverless Unavailable

Privilege to create an index or data stream. A create index request may contain aliases to be added to the index once created. In that case the request requires the manage privilege as well, on both the index and the aliases names.

cross_cluster_replication Serverless Unavailable

Privileges to perform cross-cluster replication for indices located on remote clusters configured with the API key based model. This privilege should only be used for the privileges field of remote indices privileges.

cross_cluster_replication_internal Serverless Unavailable

Privileges to perform supporting actions for cross-cluster replication from remote clusters configured with the API key based model.

Note

This privilege must not be directly granted. It is used internally by Create Cross-Cluster API key and Update Cross-Cluster API key to manage cross-cluster API keys.

delete

Privilege to delete documents.

delete_index

Privilege to delete an index or data stream.

index

Privilege to index and update documents.

Important

Starting from 8.0, this privilege no longer grants the permission to update index mappings. In earlier versions, it implicitly permitted index mapping updates (excluding data stream mappings) via the updating mapping API or through dynamic field mapping. Mapping update capabilities will be fully removed in a future major release.

maintenance

Permits refresh, flush, synced flush and force merge index administration operations. No privilege to read or write index data or otherwise manage the index.

manage

All monitor privileges plus index and data stream administration (aliases, analyze, cache clear, close, delete, exists, flush, mapping, open, field capabilities, force merge, refresh, settings, search shards, validate query).

manage_data_stream_lifecycle

All Data stream lifecycle operations relating to reading and managing the built-in lifecycle of a data stream. This includes operations such as adding and removing a lifecycle from a data stream.

manage_failure_store Stack 9.1.0

All monitor privileges plus index and data stream administration limited to failure stores only. Applies only to data streams when accessed through the index component selector syntax.

manage_follow_index Serverless Unavailable

All actions that are required to manage the lifecycle of a follower index, which includes creating a follower index, closing it, and converting it to a regular index. This privilege is necessary only on clusters that contain follower indices.

manage_ilm Serverless Unavailable

All index lifecycle management operations relating to managing the execution of policies of an index or data stream. This includes operations such as retrying policies and removing a policy from an index or data stream.

manage_leader_index Serverless Unavailable

All actions that are required to manage the lifecycle of a leader index, which includes forgetting a follower. This privilege is necessary only on clusters that contain leader indices.

monitor

All actions that are required for monitoring (recovery, segments info, index stats and status).

read

Read-only access to actions (count, explain, get, mget, get indexed scripts, more like this, multi percolate/search/termvector, percolate, scroll, clear_scroll, search, suggest, tv).

read_cross_cluster Serverless Unavailable

Read-only access to the search action from a remote cluster.

read_failure_store Stack 9.1.0

Read-only access to actions performed on a data stream's failure store. Required for access to failure store data (count, explain, get, mget, get indexed scripts, more like this, multi percolate/search/termvector, percolate, scroll, clear_scroll, search, suggest, tv). Applies only to data streams when accessed through the index component selector syntax.

view_index_metadata

Read-only access to index and data stream metadata (aliases, exists, field capabilities, field mappings, get index, get data stream, ilm explain, mappings, search shards, settings, validate query). This privilege is available for use primarily by Kibana users.

write

Privilege to perform all write operations to documents, which includes the permission to index, update, and delete documents as well as performing bulk operations, while also allowing to dynamically update the index mapping.

Important

Starting from 8.0, this privilege no longer grants the permission to update index mappings. In earlier versions, it implicitly permitted index mapping updates (excluding data stream mappings) via the updating mapping API or through dynamic field mapping. Mapping update capabilities will be fully removed in a future major release.

Serverless Unavailable

The run_as permission enables an authenticated user to submit requests on behalf of another user. The value can be a user name or a comma-separated list of user names. (You can also specify users as an array of strings or a YAML sequence.) For more information, see Submitting requests on behalf of other users.

Application privileges are managed within Elasticsearch and can be retrieved with the has privileges API and the get application privileges API. They do not, however, grant access to any actions or resources within Elasticsearch. Their purpose is to enable applications to represent and store their own privilege models within Elasticsearch roles.

To create application privileges, use the add application privileges API. You can then associate these application privileges with roles, as described in Defining roles.