You're viewing version 2.5 of the OpenSearch documentation. This version is no longer maintained. For the latest version, see the current documentation. For information about OpenSearch version maintenance, see Release Schedule and Maintenance Policy.

Anomaly detection security

You can use the Security plugin with anomaly detection in OpenSearch to limit non-admin users to specific actions. For example, you might want some users to only be able to create, update, or delete detectors, while others to only view detectors.

All anomaly detection indices are protected as system indices. Only a super admin user or an admin user with a TLS certificate can access system indices. For more information, see System indices.

Security for anomaly detection works the same as security for alerting.

Basic permissions

As an admin user, you can use the Security plugin to assign specific permissions to users based on which APIs they need access to. For a list of supported APIs, see Anomaly detection API.

The Security plugin has two built-in roles that cover most anomaly detection use cases: anomaly_full_access and anomaly_read_access. For descriptions of each, see Predefined roles.

If these roles don’t meet your needs, mix and match individual anomaly detection permissions to suit your use case. Each action corresponds to an operation in the REST API. For example, the cluster:admin/opensearch/ad/detector/delete permission lets you delete detectors.

A note on alerts and fine-grained access control

When a trigger generates an alert, the detector and monitor configurations, the alert itself, and any notification that is sent to a channel may include metadata describing the index being queried. By design, the plugin must extract the data and store it as metadata outside of the index. Document-level security (DLS) and field-level security (FLS) access controls are designed to protect the data in the index. But once the data is stored outside the index as metadata, users with access to the detector and monitor configurations, alerts, and their notifications will be able to view this metadata and possibly infer the contents and quality of data in the index, which would otherwise be concealed by DLS and FLS access control.

To reduce the chances of unintended users viewing metadata that could describe an index, we recommend that administrators enable role-based access control and keep these kinds of design elements in mind when assigning permissions to the intended group of users. See Limit access by backend role for details.

(Advanced) Limit access by backend role

Use backend roles to configure fine-grained access to individual detectors based on roles. For example, users of different departments in an organization can view detectors owned by their own department.

First, make sure your users have the appropriate backend roles. Backend roles usually come from an LDAP server or SAML provider, but if you use the internal user database, you can use the REST API to add them manually.

Next, enable the following setting:

PUT _cluster/settings
{
  "transient": {
    "plugins.anomaly_detection.filter_by_backend_roles": "true"
  }
}

Now when users view anomaly detection resources in OpenSearch Dashboards (or make REST API calls), they only see detectors created by users who share at least one backend role. For example, consider two users: alice and bob.

alice has an analyst backend role:

PUT _plugins/_security/api/internalusers/alice
{
  "password": "alice",
  "backend_roles": [
    "analyst"
  ],
  "attributes": {}
}

bob has a human-resources backend role:

PUT _plugins/_security/api/internalusers/bob
{
  "password": "bob",
  "backend_roles": [
    "human-resources"
  ],
  "attributes": {}
}

Both alice and bob have full access to anomaly detection:

PUT _plugins/_security/api/rolesmapping/anomaly_full_access
{
  "backend_roles": [],
  "hosts": [],
  "users": [
    "alice",
    "bob"
  ]
}

Because they have different backend roles, alice and bob cannot view each other’s detectors or their results.