IMPORTANT: No additional bug fixes or documentation updates will be released for this version. For the latest information, see the current release documentation.

« Metricbeat features that require authorization Grant users access to Metricbeat indices »

› › ›

Configure authentication credentials

edit

Configure authentication credentials

edit

When sending data to a secured cluster through the elasticsearch output, Metricbeat must either provide basic authentication credentials or present a client certificate.

To configure authentication credentials for Metricbeat:

Create a writer role that has the following privileges:
- Cluster: manage_index_templates and monitor
- Index: write and create_index on the Metricbeat indices
You can create roles from the Management / Roles UI in Kibana or through the role API. For example, the following request creates a role named metricbeat_writer:
```
POST _xpack/security/role/metricbeat_writer
{
  "cluster": ["manage_index_templates","monitor"],
  "indices": [
    {
      "names": [ "metricbeat-*" ], 
      "privileges": ["write","create_index"]
    }
  ]
}
```
If you use a custom Metricbeat index pattern, specify that pattern instead of the default metricbeat-* pattern.
If you plan to use index lifecycle management, create a role that has the following privileges. These privileges are required to load index lifecycle policies and create and manage rollover indices:
- Cluster: manage_ilm
- Index: write, create_index, manage, and manage_ilm on the Metricbeat indices
```
POST _xpack/security/role/metricbeat_ilm
{
  "cluster": ["manage_ilm"],
  "indices": [
    {
      "names": [ "metricbeat-*","shrink-metricbeat-*"],
      "privileges": ["write","create_index","manage","manage_ilm"]
    }
  ]
}
```
Assign the writer role to the user that Metricbeat will use to connect to Elasticsearch. Make sure you also assign any roles that are required for specific features. For the list of features and required roles, see Metricbeat features that require authorization.
1. To authenticate as a native user, create a user for Metricbeat to use internally and assign it the writer role, plus any other roles that are needed.
  
  You can create users from the Management / Users UI in Kibana or through the user API. For example, following request creates a user named metricbeat_internal that has the metricbeat_writer and kibana_user roles:
```
POST /_xpack/security/user/metricbeat_internal
{
  "password" : "YOUR_PASSWORD",
  "roles" : [ "metricbeat_writer","kibana_user"],
  "full_name" : "Internal Metricbeat User"
}
```
2. To use PKI authentication, assign the writer role, plus any other roles that are needed, in the role_mapping.yml configuration file. Specify the user by the distinguished name that appears in its certificate:
```
metricbeat_writer:
  - "cn=Internal Metricbeat User,ou=example,o=com"
kibana_user:
  - "cn=Internal Metricbeat User,ou=example,o=com"
```
  For more information, see Using Role Mapping Files.

In the Metricbeat configuration file, specify authentication credentials for the elasticsearch output:

To use basic authentication, configure the username and password settings. For example, the following Metricbeat output configuration uses the native metricbeat_internal user to connect to Elasticsearch:
```
output.elasticsearch:
  hosts: ["localhost:9200"]
  username: "metricbeat_internal" 
  password: "YOUR_PASSWORD" 
```
You created this user earlier.

The example shows a hard-coded password, but you should store sensitive values in the secrets keystore.

To use PKI authentication, configure the certificate and key settings:

output.elasticsearch:
  hosts: ["localhost:9200"]
  ssl.certificate: "/etc/pki/client/cert.pem" 
  ssl.key: "/etc/pki/client/cert.key"

The distinguished name (DN) in the certificate must be mapped to the metricbeat_writer and kibana_user roles in the role_mapping.yml configuration file on each node in the Elasticsearch cluster.

« Metricbeat features that require authorization Grant users access to Metricbeat indices »

	You created this user earlier.
	The example shows a hard-coded password, but you should store sensitive values in the secrets keystore.