Configure the Elasticsearch output
editConfigure the Elasticsearch output
editThe Elasticsearch output sends events directly to Elasticsearch using the Elasticsearch HTTP API.
Example configuration:
When sending data to a secured cluster through the elasticsearch
output, Metricbeat can use any of the following authentication methods:
- Basic authentication credentials (username and password).
- Token-based (API key) authentication.
- Public Key Infrastructure (PKI) certificates.
Basic authentication:
output.elasticsearch: hosts: ["https://myEShost:9200"] username: "metricbeat_writer" password: "YOUR_PASSWORD"
API key authentication:
output.elasticsearch: hosts: ["https://myEShost:9200"] api_key: "ZCV7VnwBgnX0T19fN8Qe:KnR6yE41RrSowb0kQ0HWoA"
PKI certificate authentication:
output.elasticsearch: hosts: ["https://myEShost:9200"] ssl.certificate: "/etc/pki/client/cert.pem" ssl.key: "/etc/pki/client/cert.key"
See Secure communication with Elasticsearch for details on each authentication method.
Compatibility
editThis output works with all compatible versions of Elasticsearch. See the Elastic Support Matrix.
Optionally, you can set Metricbeat to only connect to instances that are at least on the
same version as the Beat. The check can be enabled by setting output.elasticsearch.allow_older_versions
to false
. Leaving the setting at it’s default value of true
avoids an issue where Metricbeat cannot connect to Elasticsearch after having been upgraded to a version higher than the Elastic Stack.
Configuration options
editYou can specify the following options in the elasticsearch
section of the metricbeat.yml
config file:
enabled
editThe enabled config is a boolean setting to enable or disable the output. If set
to false
, the output is disabled.
The default value is true
.
hosts
editThe list of Elasticsearch nodes to connect to. The events are distributed to
these nodes in round robin order. If one node becomes unreachable, the event is
automatically sent to another node. Each Elasticsearch node can be defined as a URL
or IP:PORT
.
For example: http://192.15.3.2
, https://es.found.io:9230
or 192.24.3.2:9300
.
If no port is specified, 9200
is used.
output.elasticsearch: hosts: ["10.45.3.2:9220", "10.45.3.1:9230"] protocol: https path: /elasticsearch
In the previous example, the Elasticsearch nodes are available at https://10.45.3.2:9220/elasticsearch
and
https://10.45.3.1:9230/elasticsearch
.
compression_level
editThe gzip compression level. Setting this value to 0
disables compression.
The compression level must be in the range of 1
(best speed) to 9
(best compression).
Increasing the compression level will reduce the network usage but will increase the cpu usage.
The default value is 1
.
escape_html
editConfigure escaping of HTML in strings. Set to true
to enable escaping.
The default value is false
.
worker
or workers
editThe number of workers per configured host publishing events to Elasticsearch. This is best used with load balancing mode enabled. Example: If you have 2 hosts and 3 workers, in total 6 workers are started (3 for each host).
The default value is 1
.
loadbalance
editWhen loadbalance: true
is set, Metricbeat connects to all configured
hosts and sends data through all connections in parallel. If a connection
fails, data is sent to the remaining hosts until it can be reestablished.
Data will still be sent as long as Metricbeat can connect to at least
one of its configured hosts.
When loadbalance: false
is set, Metricbeat sends data to a single host
at a time. The target host is chosen at random from the list of configured
hosts, and all data is sent to that target until the connection fails, when
a new target is selected. Data will still be sent as long as Metricbeat
can connect to at least one of its configured hosts.
The default value is true
.
output.elasticsearch: hosts: ["localhost:9200", "localhost:9201"] loadbalance: true
api_key
editInstead of using a username and password, you can use API keys to secure communication
with Elasticsearch. The value must be the ID of the API key and the API key joined by a colon: id:api_key
.
See Grant access using API keys for more information.
username
editThe basic authentication username for connecting to Elasticsearch.
This user needs the privileges required to publish events to Elasticsearch. To create a user like this, see Create a publishing user.
password
editThe basic authentication password for connecting to Elasticsearch.
parameters
editDictionary of HTTP parameters to pass within the url with index operations.
protocol
editThe name of the protocol Elasticsearch is reachable on. The options are:
http
or https
. The default is http
. However, if you specify a URL for
hosts
, the value of protocol
is overridden by whatever scheme you
specify in the URL.
path
editAn HTTP path prefix that is prepended to the HTTP API calls. This is useful for the cases where Elasticsearch listens behind an HTTP reverse proxy that exports the API under a custom prefix.
headers
editCustom HTTP headers to add to each request created by the Elasticsearch output. Example:
output.elasticsearch.headers: X-My-Header: Header contents
It is possible to specify multiple header values for the same header name by separating them with a comma.
proxy_disable
editIf set to true
all proxy settings, including HTTP_PROXY
and HTTPS_PROXY
variables are ignored.
proxy_url
editThe URL of the proxy to use when connecting to the Elasticsearch servers. The value must be a complete URL. If a value is not specified through the configuration file then proxy environment variables are used. See the Go documentation for more information about the environment variables.
proxy_headers
editAdditional headers to send to proxies during CONNECT requests.
index
editThe index name to write events to when you’re using daily indices. The default is
"metricbeat-%{[agent.version]}-%{+yyyy.MM.dd}"
, for example,
"metricbeat-8.14.3-2024-09-06"
. If you change this setting, you also
need to configure the setup.template.name
and setup.template.pattern
options
(see Elasticsearch index template).
If you are using the pre-built Kibana
dashboards, you also need to set the setup.dashboards.index
option (see
Kibana dashboards).
When index lifecycle management (ILM) is enabled, the default index
is
"metricbeat-%{[agent.version]}-%{+yyyy.MM.dd}-%{index_num}"
, for example,
"metricbeat-8.14.3-2024-09-06-000001"
. Custom index
settings are ignored
when ILM is enabled. If you’re sending events to a cluster that supports index
lifecycle management, see Index lifecycle management (ILM) to learn how to change the index name.
You can set the index dynamically by using a format string to access any event
field. For example, this configuration uses a custom field, fields.log_type
,
to set the index:
output.elasticsearch: hosts: ["http://localhost:9200"] index: "%{[fields.log_type]}-%{[agent.version]}-%{+yyyy.MM.dd}"
With this configuration, all events with log_type: normal
are sent to an
index named normal-8.14.3-2024-09-06
, and all events with
log_type: critical
are sent to an index named
critical-8.14.3-2024-09-06
.
To learn how to add custom fields to events, see the
fields
option.
See the indices
setting for other ways to set the index
dynamically.
indices
editAn array of index selector rules. Each rule specifies the index to use for
events that match the rule. During publishing, Metricbeat uses the first
matching rule in the array. Rules can contain conditionals, format string-based
fields, and name mappings. If the indices
setting is missing or no rule
matches, the index
setting is used.
Similar to index
, defining custom indices
will disable Index lifecycle management (ILM).
Rule settings:
-
index
-
The index format string to use. If this string contains field
references, such as
%{[fields.name]}
, the fields must exist, or the rule fails. -
mappings
-
A dictionary that takes the value returned by
index
and maps it to a new name. -
default
-
The default string value to use if
mappings
does not find a match. -
when
- A condition that must succeed in order to execute the current rule. All the conditions supported by processors are also supported here.
The following example sets the index based on whether the message
field
contains the specified string:
output.elasticsearch: hosts: ["http://localhost:9200"] indices: - index: "warning-%{[agent.version]}-%{+yyyy.MM.dd}" when.contains: message: "WARN" - index: "error-%{[agent.version]}-%{+yyyy.MM.dd}" when.contains: message: "ERR"
This configuration results in indices named warning-8.14.3-2024-09-06
and error-8.14.3-2024-09-06
(plus the default index if no matches are
found).
The following example sets the index by taking the name returned by the index
format string and mapping it to a new name that’s used for the index:
output.elasticsearch: hosts: ["http://localhost:9200"] indices: - index: "%{[fields.log_type]}" mappings: critical: "sev1" normal: "sev2" default: "sev3"
This configuration results in indices named sev1
, sev2
, and sev3
.
The mappings
setting simplifies the configuration, but is limited to string
values. You cannot specify format strings within the mapping pairs.
ilm
editConfiguration options for index lifecycle management.
See Index lifecycle management (ILM) for more information.
pipeline
editA format string value that specifies the ingest pipeline to write events to.
output.elasticsearch: hosts: ["http://localhost:9200"] pipeline: my_pipeline_id
For more information, see Parse data using an ingest pipeline.
You can set the ingest pipeline dynamically by using a format string to
access any event field. For example, this configuration uses a custom field,
fields.log_type
, to set the pipeline for each event:
output.elasticsearch: hosts: ["http://localhost:9200"] pipeline: "%{[fields.log_type]}_pipeline"
With this configuration, all events with log_type: normal
are sent to a pipeline
named normal_pipeline
, and all events with log_type: critical
are sent to a
pipeline named critical_pipeline
.
To learn how to add custom fields to events, see the
fields
option.
See the pipelines
setting for other ways to set the
ingest pipeline dynamically.
pipelines
editAn array of pipeline selector rules. Each rule specifies the ingest
pipeline to use for events that match the rule. During publishing, Metricbeat
uses the first matching rule in the array. Rules can contain conditionals,
format string-based fields, and name mappings. If the pipelines
setting is
missing or no rule matches, the pipeline
setting is
used.
Rule settings:
-
pipeline
-
The pipeline format string to use. If this string contains field
references, such as
%{[fields.name]}
, the fields must exist, or the rule fails. -
mappings
-
A dictionary that takes the value returned by
pipeline
and maps it to a new name. -
default
-
The default string value to use if
mappings
does not find a match. -
when
- A condition that must succeed in order to execute the current rule. All the conditions supported by processors are also supported here.
The following example sends events to a specific pipeline based on whether the
message
field contains the specified string:
output.elasticsearch: hosts: ["http://localhost:9200"] pipelines: - pipeline: "warning_pipeline" when.contains: message: "WARN" - pipeline: "error_pipeline" when.contains: message: "ERR"
The following example sets the pipeline by taking the name returned by the
pipeline
format string and mapping it to a new name that’s used for the
pipeline:
output.elasticsearch: hosts: ["http://localhost:9200"] pipelines: - pipeline: "%{[fields.log_type]}" mappings: critical: "sev1_pipeline" normal: "sev2_pipeline" default: "sev3_pipeline"
With this configuration, all events with log_type: critical
are sent to
sev1_pipeline
, all events with log_type: normal
are sent to a
sev2_pipeline
, and all other events are sent to sev3_pipeline
.
For more information about ingest pipelines, see Parse data using an ingest pipeline.
max_retries
editThe number of times to retry publishing an event after a publishing failure. After the specified number of retries, the events are typically dropped.
Set max_retries
to a value less than 0 to retry until all events are published.
The default is 3.
bulk_max_size
editThe maximum number of events to bulk in a single Elasticsearch bulk API index request. The default is 1600.
Events can be collected into batches. Metricbeat will split batches read from the queue which are
larger than bulk_max_size
into multiple batches.
Specifying a larger batch size can improve performance by lowering the overhead of sending events. However big batch sizes can also increase processing times, which might result in API errors, killed connections, timed-out publishing requests, and, ultimately, lower throughput.
Setting bulk_max_size
to values less than or equal to 0 disables the
splitting of batches. When splitting is disabled, the queue decides on the
number of events to be contained in a batch.
backoff.init
editThe number of seconds to wait before trying to reconnect to Elasticsearch after
a network error. After waiting backoff.init
seconds, Metricbeat tries to
reconnect. If the attempt fails, the backoff timer is increased exponentially up
to backoff.max
. After a successful connection, the backoff timer is reset. The
default is 1s
.
backoff.max
editThe maximum number of seconds to wait before attempting to connect to
Elasticsearch after a network error. The default is 60s
.
idle_connection_timeout
editThe maximum amount of time an idle connection will remain idle before closing itself. Zero means no limit. The format is a Go language duration (example 60s is 60 seconds). The default is 3s.
timeout
editThe http request timeout in seconds for the Elasticsearch request. The default is 90.
allow_older_versions
editBy default, Metricbeat expects the Elasticsearch instance to be on the same or newer version to provide optimal experience. We suggest you connect to the same version to make sure all features Metricbeat is using are available in your Elasticsearch instance.
You can disable the check for example during updating the Elastic Stack, so data collection can go on.
ssl
editConfiguration options for SSL parameters like the certificate authority to use
for HTTPS-based connections. If the ssl
section is missing, the host CAs are used for HTTPS connections to
Elasticsearch.
See the secure communication with Elasticsearch guide or SSL configuration reference for more information.
non_indexable_policy
editSpecifies the behavior when the elasticsearch cluster explicitly rejects documents, for example on mapping conflicts.
drop
editThe default behaviour, when an event is explicitly rejected by elasticsearch it is dropped.
output.elasticsearch: hosts: ["http://localhost:9200"] non_indexable_policy.drop: ~
dead_letter_index
editThis functionality is in beta and is subject to change. The design and code is less mature than official GA features and is being provided as-is with no warranties. Beta features are not subject to the support SLA of official GA features.
On an explicit rejection, this policy will retry the event in the next batch. However, the target index will change to index specified. In addition, the structure of the event will be change to the following fields:
- message
- Contains the escaped json of the original event.
- error.type
- Contains the status code
- error.message
- Contains status returned by elasticsearch, describing the reason
-
index
- The index to send rejected events to.
output.elasticsearch: hosts: ["http://localhost:9200"] non_indexable_policy.dead_letter_index: index: "my-dead-letter-index"
preset
editThe performance preset to apply to the output configuration.
output.elasticsearch: hosts: ["http://localhost:9200"] preset: balanced
Performance presets apply a set of configuration overrides based on a desired performance goal. If set, a performance preset will override other configuration flags to match the recommended settings for that preset. Valid options are:
* balanced
: good starting point for general efficiency
* throughput
: good for high data volumes, may increase cpu and memory requirements
* scale
: reduces ambient resource use in large low-throughput deployments
* latency
: minimize the time for fresh data to become visible in Elasticsearch
* custom
: apply user configuration directly with no overrides
The default if unspecified is custom
.
Presets represent current recommendations based on the intended goal; their effect may change between versions to better suit those goals. Currently the presets have the following effects:
preset | balanced | throughput | scale | latency |
---|---|---|---|---|
1600 |
1600 |
1600 |
50 |
|
1 |
4 |
1 |
1 |
|
3200 |
12800 |
3200 |
4100 |
|
1600 |
1600 |
1600 |
2050 |
|
|
|
|
|
|
1 |
1 |
1 |
1 |
|
|
|
|
|
Elasticsearch APIs
editMetricbeat will use the _bulk
API from Elasticsearch, the events are sent
in the order they arrive to the publishing pipeline, a single _bulk
request may contain events from different inputs/modules. Temporary
failures are re-tried.
The status code for each event is checked and handled as:
-
< 300
: The event is counted asevents.acked
-
409
(Conflict): The event is counted asevents.duplicates
-
429
(Too Many Requests): The event is counted asevents.toomany
-
> 399 and < 500
: Thenon_indexable_policy
is applied.