Set up machine learning features
editSet up machine learning features
editTo use the Elastic Stack machine learning features, you must have the appropriate subscription and at least one machine learning node in your cluster.
If Elastic Stack security features are enabled, you must also ensure your users have the necessary privileges. If the operator privileges feature is enabled, there are some machine learning settings that can be updated only by operator users.
The fastest way to get started with machine learning features is to start a free 14-day trial of Elasticsearch Service in the cloud.
Machine learning nodes
editTo use machine learning features, there must be at least one machine learning node in your cluster. A
machine learning node is a node that has xpack.ml.enabled
set to true
and ml
in
node.roles
.
If nodes do not have the machine learning role, they cannot run machine learning jobs. If
xpack.ml.enabled
is true
, however, they can service API requests. For more
information, see Machine learning nodes and
Machine learning settings in Elasticsearch.
Security privileges
editElasticsearch security privileges
editThe Elastic Stack security features provide roles and privileges that make it easier to control which users can manage or view machine learning objects such as jobs, datafeeds, results, and model snapshots.
If you use machine learning APIs, you must have the machine_learning_admin
or
machine_learning_user
built-in roles or the equivalent cluster privileges and
the following index privileges:
For full access:
-
❏
read
andview_index_metadata
on source indices -
❏
read
,manage
, andindex
on destination indices (for data frame analytics jobs only)
For read-only access:
-
❏
read
index privileges on source indices -
❏
read
index privileges on destination indices (for data frame analytics jobs only)
Kibana privileges
editIn Kibana, the machine learning features must be visible in your space and your source index patterns must exist in the same space as your machine learning jobs.
Kibana enables you to control access to the machine learning features within each space. You can manage your roles, privileges, and spaces in the Stack Management app in Kibana. For more information, see Security privileges and Kibana privileges.
The machine_learning_admin
and machine_learning_user
roles grant access to
the machine learning features in all Kibana spaces. Therefore, when you use Kibana, use custom
roles instead and set your Kibana privileges appropriately for each space.
For full access to the machine learning features in Kibana, you must have:
-
all
Kibana privileges for the machine learning features in the appropriate spaces -
read
andview_index_metadata
index privileges on source indices -
read
,manage
, andindex
index privileges on destination indices (for data frame analytics jobs only)
For read-only access to the machine learning features in Kibana, you must have:
-
read
Kibana privileges for the machine learning features in the appropriate spaces -
read
index privileges on source indices -
read
index privileges on destination indices (for data frame analytics jobs only)
To upload and import files in the Data Visualizer, you must have:
-
all
Kibana privileges for the Machine Learning or Discover features in the appropriate spaces. Alternatively,read
Kibana privileges for the Machine Learning feature andall
Kibana privileges for the Index Pattern Management feature -
manage_pipeline
ormanage_ingest_pipelines
cluster privileges -
create
,create_index
,manage
andread
index privileges for destination indices
You cannot limit access to specific machine learning objects in each space. If
the machine learning feature is visible in your space and you have read
or all
Kibana
privileges for the feature, you have access to all machine learning objects in that space.