Infrastructure agent configuration settings

Specifies the license key for your New Relic account. The agent uses this key to associate your server's metrics with your New Relic account. This setting is created as part of the standard installation process.

Example:

license_key: 1234567890abcdefghijklmnopqrstuvwxyz1234

Specifies whether FedRAMP endpoints should be used.

Specifies the number of logical processors available to the agent. Increasing this value helps to distribute the load between different cores.

If set to -1, the agent will try to read the environment variable GOMAXPROCS. If this variable is not set, the default value will be the total number of cores available in the host.

Since version 1.0.804 or higher, data sent from the agent is compressed by default. To disable payload compression, set payload_compression_level to 0.

Number of times the agent will retry the request to check New Relic's platform availability at startup before throwing an error.

If set to a negative value, the agent will keep checking the connection until it succeeds.

After a request has timed out, time the agent waits to retry a request to check New Relic's platform availability at startup.

Time the agent waits until a request to check New Relic's platform availability at startup is considered timed out.

Enables agent's self instrumentation.

Enables Agent Retry for Log Transmission via the embedded fluentbit logging forwarder. Integer values are for the number of intended retries. Other possible values include False to set the number of retries to infinite and no_retries to turn off the retry functionality entirely.

Sets the number of times the agent retries to connect in case that cloud detection fails.

If cloud detection fails during the initialization of the agent, the agent will retry after waiting for CloudRetryBackOffSec seconds.

Sets the interval of time the agent will wait until discarding the metadata, in seconds. After this period metadata expires and the agent will fetch it again.

Sets the interval of time the agent waits between cloud detection retries in case that cloud detection failed, in seconds.

If cloud detection fails during the initialization of the agent, it will retry for CloudMaxRetryCount times.

Specifies the cloud provider the agent is running in. When this is set up to a value different from the default, the agent will wait until it successfully acquires metadata (the instance ID) from the cloud provider before submitting any data to the backend.

With the default behavior, the agent will also try to detect the cloud provider and get the metadata but without blocking the agent.

Allowed values: "", "aws", "azure", "gcp", "alibaba"

Similar to DisableCloudMetadata, but it disables the collection of cloud metadata only for the host alias plugin.

Disables the collection of cloud metadata.

Time, in seconds, before the cached containers metadata expires and the agent needs to fetch them again.

Specifies the Docker client API version.

List of the types of file systems that the agent supports. This value needs to be a subset of the default list, and items that are not in the default list will be discarded.

List of storage devices to be ignored by the agent when gathering StorageSample data.

Example as a YAML attribute:

file_devices_ignored:
  - sda1
  - sda2

Example as an environment variable:

FILE_DEVICES_IGNORED="sda1,sda2"

Overrides the auto-generated hostname for reporting. This is useful when you have multiple hosts with the same name, since our infrastructure monitoring uses the hostname as the unique identifier for each host. Keep in mind this value is also used for the loopback address replacement on entity names.

For more information, see our documentation on how entity name resolution works.

Example:

display_name: teslaOne

When true, the full hostname is resolved by performing a reverse lookup of the host's address. Otherwise, it will be retrieved with the hostname command on Linux and from the TCP/IP parameters of the registry on Windows.

When set, this is the value that will be reported for the full hostname; otherwise, the agent will perform the normal lookup behavior.

Example:

my.custom-hostname.co.org

When set, this is the value that will be reported for the hostname; otherwise, the agent will perform the normal lookup behavior.

Example:

my.custom-hostname

Directory where the agent stores files for cache, inventory, integrations, etc.

Directory containing the configuration files of the integrations.

Each integration has its own configuration file, named by default <integration_name>-config.yml, placed in a predefined location from which the agent loads on initialization.

Specifies a custom path to install integrations, which allows to install them outside the agent_dir. It has priority when the agent is looking for installed integrations.

A list of environment variables that will be passed to all integrations. If an integration already has an existing configuration option with the same name, then the environment variable takes precedence.

Starting with infrastructure agent 1.24.1 passthrough_environment supports regex for variable names.

Example as a YAML attribute (inside the agent's configuration file, located by default in /etc/newrelic-infra.yml):

passthrough_environment:
  - HOST
  - PORT
  - NRIA_.*

Example as an environment variable:

NRIA_PASSTHROUGH_ENVIRONMENT="HOST,PORT,NRIA_.*"

For Windows hosts: PATHEXT is an MS Windows environment variable. The function is to determine which file extensions mark files that are executable from every command line. This will resolve any errors reporting "... is not recognized as the name of a cmdlet, function, script file..." for common Flex integrations:

passthrough_environment:
  - PATHEXT

The agent enables loopback-address replacement on the entity name (and therefore key) automatically for version 3 of the integration protocol. If you are using version 2 of the protocol and want this behavior, enable the entityname_integrations_v2_update option.

By setting this configuration parameter to true the agent will open an HTTP port (by default, 8001) to receive data from the New Relic StatsD backend.

By setting this value, the agent will start listening on the HTTPServerPort to receive data from the New Relic StatsD backend.

Sets the port for the http server to receive data from the New Relic StatsD backend.

Starts the process of deleting entities that haven't reported information during this interval.

Valid time units: s (seconds), m (minutes), and h (hour).

Example:

1h

If you have a large number of processes on the host, or a large volume of data collected across your integrations, the agent queue can overrun. Increase the queue depth to prevent error messages resulting from full queues.

Defines the temp directory used as persisting storage for the integrations SDK synchronization.

Use this setting if you're instrumenting the host from a source other than the infrastructure agent (for example, an OpenTelemetry Collector or Prometheus node exporter) and you want to keep using the infrastructure agent on-host integrations to monitor other infrastructure services. When enabled, the agent reports host inventory and integrations telemetry (event metrics and inventory) decorated with host metadata, but host metrics (CPU, memory, disk, network, processes) are disabled.

If the cached inventory becomes older than this value (for example, because the agent is offline), the agent automatically removes and recreates the delta store.

When set to true, enables the inventory asynchronous processing allowing larger inventory payloads to be processed.

The list of inventory paths ignored by the agent.

Example as a YAML attribute:

ignored_inventory:
  - files/config/stuff.bar
  - files/config/stuff.foo

Example as an environment variable:

NRIA_IGNORED_INVENTORY="files/config/stuff.bar,files/config/stuff.foo"

Location on Linux where the pid file of the agent process is created. It is used at startup to ensure that no other instances of the agent are running.

When true, formulation of the host virtual memory considers SReclaimable as available memory; otherwise SReclaimable will be considered part of the used memory.

Log is a key-value map used to configure the infrastructure agent logger. Example as a YAML attribute:

log:
  file: '/var/log/newrelic-infra/newrelic-infra.log'
  level: debug
  forward: true
  stdout: false

Example as an environment variable:

NRIA_LOG_FILE='/var/log/newrelic-infra/newrelic-infra.log'
NRIA_LOG_LEVEL='smart'
NRIA_LOG_FORWARD='false'
NRIA_LOG_STDOUT='true'

log:
  level: info
  file: /var/log/newrelic-infra/newrelic-infra.log
  rotate:
    max_size_mb: 1000                        # Required
    max_files: 5                             # Optional
    compression_enabled: true                # Optional
    file_pattern: YYYY-MM-DD_hh-mm-ss.log    # Optional

Custom attributes are key-value pairs (similar to tags in other tools) used to annotate the data from the infrastructure agent. You can use this metadata to filter your entities, group your results, and annotate your data. For example, you might indicate a machine's environment (staging or production), the service a machine hosts (login service, for example), or the team responsible for that machine.

Use a list of custom attributes to annotate the data from this agent instance. Separate keys and values with colons :, as in KEY: VALUE, and separate each key-value pair with a line break. Keys can be any valid YAML except slashes /. Values can be any YAML string, including spaces. Starting the key with label. will make sure it remains as is, even if that host might run in different cloud environments, where the cloud tags or cloud labels might interact with it in the combined decoration of tags. In contrast, custom_labels might be prefixed with gcp. or azure. when running in those environments and when combining data from the infrastructure agent with cloud monitoring together.

Example as a YAML attribute:

custom_attributes:
  label.environment: production
  label.service: login service
  label.team: alpha-team

Example as an environment variable:

NRIA_CUSTOM_ATTRIBUTES='{"label.customAttribute_1":"SOME_ATTRIBUTE","label.customAttribute_2": "SOME_ATTRIBUTE_2"}'

NRQL example filtering by custom attribute:

FROM SystemSample SELECT * WHERE label.environment = 'production'

Enables the sending of process metrics to New Relic.

By default, the infrastructure agent doesn't send data about the operating system's processes. The agent still collects such data, unless metrics_process_sample_rate is set to -1.

To report metric data about all the operating system's processes, set enable_process_metrics to true. To disable, set to false.

Sending all process data could increase the volume of data sent to New Relic. To fine-tune which processes you want to monitor, configure include_matching_metrics or exclude_matching_metrics.

By default, processes using low memory are excluded from being sampled. For more information, see disable-zero-mem-process-filter.

When enabled, the agent decorates process samples with container information if the process is executed in a container.

This information includes the container ID, container name, image ID, image name, and container labels.

You can control how much data is sent to New Relic by configuring include_matching_metrics, which allows you to restrict the transmission of metric data based on the values of metric attributes.

You include metric data by defining literal or partial values for any of the attributes of the metric. For example, you can choose to send the host.process.cpuPercent of all processes whose process.name match the ^java regular expression.

In this example, we include process metrics using executable files and names:

include_matching_metrics:          # You can combine attributes from different metrics
  process.name:
    - regex "^java"                # Include all processes starting with "java"
  process.executable:
    - "/usr/bin/python2"           # Include the Python 2.x executable
    - regex "\\System32\\svchost"  # Include all svchost executables

If you need to include command line arguments in any of the values, set strip_command_line to false (the infrastructure agents removes CLI arguments by default to prevent secrets from leaking).

If both include_matching_metrics and exclude_matching_metrics are present for the same item, the inclusion takes precedence.

To configure include_matching_metrics as an environment variable for the Kubernetes integration, add it in the manifest inside the env: object:

env:
  - name: NRIA_INCLUDE_MATCHING_METRICS
    value: |
      process.name:
        - regex "^java"
      process.executable:
        - "/usr/bin/python2"
        - regex "\\System32\\svchost"

Default

You can control how much data is sent to New Relic by configuring exclude_matching_metrics, which allows you to restrict the transmission of metric data based on the values of metric attributes.

You exclude metric data by defining literal or partial values for any of the attributes of the metric. For example, you can choose not to send the host.process.cpuPercent of all processes whose process.name match the ^java regular expression.

In this example, we exclude process metrics using executable files and names:

exclude_matching_metrics:          # You can combine attributes from different metrics
  process.name:
    - regex "^java"                # Exclude all processes starting with "java"
  process.executable:
    - "/usr/bin/python2"           # Exclude the Python 2.x executable
    - regex "\\System32\\svchost"  # Exclude all svchost executables

To configure exclude_matching_metrics as an environment variable for the Kubernetes integration, add it in the manifest inside the env: object:

env:
  - name: NRIA_EXCLUDE_MATCHING_METRICS
    value: |
      process.name:
        - regex "^java"
      process.executable:
        - "/usr/bin/python2"
        - regex "\\System32\\svchost"

Default

You can use the network interface filters configuration to hide unused or uninteresting network interfaces from the infrastructure agent. This helps reduce resource usage, work, and noise in your data.

The configuration uses a simple pattern-matching mechanism that can look for interfaces that start with a specific sequence of letters or numbers following either pattern:

• {name}[other characters], where you specify the name using the prefix option

• [number]{name}[other characters], where you specify the name using the index-1 option

  New Relic infrastructure implements a curated default list of filters, available for both Linux and Windows, that you can modify.

  YML option name	Environment variable	Type	Default	Version
  network_interface_filters	not supported	map[string][]string		1.0.220

  Linux

  Default network interface filters for Linux:

  • Network interfaces that start with dummy, lo, vmnet, sit, tun, tap, or veth

  • Network interfaces that contain tun or tap

    The following example (added to your configuration file) overrides the default filters. This will ignore network interfaces that start with dummy or lo , or contain tun preceded by a sequence of numbers, and followed by other characters:

    network_interface_filters:
      prefix:
        - dummy
        - lo
      index-1:
        - tun

    Copy

  Windows

  Default network interface filters for Windows:

  • Network interfaces that start with Loop, isatap, or Local

    The following example (added to your configuration file) overrides the default filters. This will ignore network interfaces that start with Loop:

    network_interface_filters:
      prefix:
        - Loop

    Copy

The ZeroRSSFilter excludes processes that are not using memory from being sampled. Disable the filter so that the agent includes these processes in the ProcessSample.

To disable all the plugins, set this option to true.

Sampling period for the CloudSecurityGroups plugin, in seconds. The minimum value is 30. To disable it, set it to -1.

Sampling period for the Daemontoolsplugin, in seconds. The minimum value is a 10. To disable it, set it to -1.

Sampling period for the Dpkg plugin, in seconds. The minimum value is 30. To disable it, set it to -1. If the parameter is not explicitly set in the config file, it can be disabled by setting DisableAllPlugins to true.

Sampling period for the Facter plugin, in seconds. The minimum value is 30. To disable it, set it to -1.

Sampling period for the KernelModules plugin, in seconds. The minimum value is 10. To disable it, set it to -1.

Sampling period for the NetworkInterface plugin, in seconds. The minimum value is 30. To disable it, set it to -1.

Sampling period for the Rpm plugin, in seconds. The minimum value is 30. To disable it, set it to -1.

Sampling period for the SELinux plugin, in seconds. The minimum value is 30. To disable it, set it to -1. This option is ignored if SelinuxEnableSemodule is set to false. For more information, see our troubleshooting doc on disabling the SELinux module.

Sampling period for the Sshd plugin, in seconds. The minimum value is 10. To disable it, set it to -1.

Sampling period for the Supervisor plugin, in seconds. The minimum value is 10. To disable it, set it to -1.

Sampling period for the Sysctl plugin, in seconds. The minimum value is 30. To disable it, set it to -1.

Sampling period for the Systemd plugin, in seconds. The minimum value is 10. To disable it, set it to -1.

Sampling period for the sysv plugin, in seconds. The minimum value is 10. To disable it, set it to -1.

Sampling period for the Upstart plugin, in seconds. The minimum value is 10. To disable it, set it to -1.

Sampling period for the Users plugin, in seconds. The minimum value is 10. To disable it, set it to -1.

Location of the supervisor socket.

Sets the HOME environment variable for Puppet's Facter. If not defined, it defaults to the current user's home directory.

Your system may have firewall rules that require the agent to use a proxy to communicate with New Relic. If so, set the proxy URL in the form https://user:password@hostname:port. It can be HTTP or HTTPS.

Example:

https://proxy_user:access_10@proxy_01:1080

When set to true, the HTTPS_PROXY and HTTP_PROXY environment variables are ignored. This is useful when the agent needs to connect directly to the metrics collector and skip the existing system proxy.

If the HTTPS_PROXY option references to a proxy with self-signed certificates, this option specifies the path to the the directory where the proxy certificate is available. The certificates in the directory must end with the .pem extension.

If the HTTPS_PROXY option references to a proxy with self-signed certificates, this option specifies the path to the certificate file.

If set to true, when the proxy is configured to use an HTTPS connection, it will only work:

• If the HTTPS proxy has certificates from a valid Certificate Authority.

• If the ca_bundle_file or ca_bundle_dir configuration properties contain the HTTPS proxy certificates.

  YML option name	Environment variable	Type	Default	Version
  proxy_validate_certificates	NRIA_PROXY_VALIDATE_CERTIFICATES	boolean	false	1.3.0

Sends the following proxy configuration information as inventory:

• HTTPS_PROXY

• HTTP_PROXY

• proxy ca_bundle_dir

• ca_bundle_file

• ignore_system_proxy

• proxy_validate_certificates

  YML option name	Environment variable	Type	Default	Version
  proxy_config_plugin	NRIA_PROXY_CONFIG_PLUGIN	boolean	true	1.3.0

Sample rate of network samples, in seconds. Minimum value is 10. To disable it, set it to -1.

Sample rate of process samples, in seconds. Minimum value is 20. To disable process samples entirely, set metrics_process_sample_rate to -1.

Sample rate of storage samples, in seconds. Minimum value is 5. To disable it, set it to -1.

Sample rate of system samples, in seconds. Minimum value is 5. To disable it, set it to -1.

Sample rate of NFS storage samples, in seconds. Minimum value is 5. To disable it, set it to -1.

Detailed NFS metrics. When enabled, the agent will provide a complete list of NFS metrics.

Flag to enable/disable the NTP Offset metric.

Defines list of NTP Hosts to query. When multiple hosts are defined, all of them will be queried and the reported metrics will be the average value between them.

Defines the interval in minutes to request NTP Offset from the provided pool.

Defines the timeout in seconds for the requests made to the NTP hosts.

Get versions of policy modules installed using SEModule. If disabled, the SELinux plugin will only retrieve the status using SEStatus.

When true, the agent removes the command arguments from the commandLine attribute of the ProcessSample.

Enables the status server on the agent providing local health information. The available endpoints are:

Main status endpoint example (with no errors):

$
curl http://localhost:8003/v1/status
{
  "checks": {
    "endpoints": [
      {
        "url": "https://infrastructure-command-api.newrelic.com/agent_commands/v1/commands",
        "reachable": true
      },
      {
        "url": "https://infra-api.newrelic.com/infra/v2/metrics",
        "reachable": true
      },
      {
        "url": "https://identity-api.newrelic.com/identity/v1",
        "reachable": true
      },
      {
        "url": "https://infra-api.newrelic.com/inventory",
        "reachable": true
      }
    ]
  },
  "config": {
    "reachability_timeout": "10s"
  }
}

Main status endpoint (with errors):

$
curl http://localhost:8003/v1/status
{
  "checks": {
    "endpoints": [
      {
        "url": "https://staging-infra-api.newrelic.com/infra/v2/metrics",
        "reachable": false,
        "error": "endpoint check timeout exceeded"
      },
      {
        "url": "https://infra-api.newrelic.com/infra/v2/metrics",
        "reachable": true
      },
      {
        "url": "https://identity-api.newrelic.com/identity/v1",
        "reachable": true
      },
      {
        "url": "https://infra-api.newrelic.com/inventory",
        "reachable": true
      }
    ]
  },
  "config": {
    "reachability_timeout": "10s"
  }
}

Errors endpoint example:

$
curl http://localhost:18003/v1/status/errors
{
  "checks": {
    "endpoints": [
      {
        "url": "https://staging-infra-api.newrelic.com/infra/v2/metrics",
        "reachable": false,
        "error": "endpoint check timeout exceeded"
      }
    ]
  },
  "config": {
    "reachability_timeout": "10s"
  }
}

This is similar to the main status endpoint but filtering those with errors only.

Entity endpoint example:

$
curl http://localhost:8003/v1/status/entity
{
  "guid":"MMMMNjI0NjR8SU5GUkF8TkF8ODIwMDg3MDc0ODE0MTUwNTMy",
  "key":"your-host-name"
}

Returns information about the agent/host entity. A response status code 204 ("No Content") will be returned when the agent still has no information about the agent/host entity. Therefore, it may take several requests to until the agent provides entity data.

Defines the port for the status server endpoint.

Sampling period for the Windows services plugin, in seconds. The minimum value is 10. To disable it, set it to -1.

Sampling period for the Windows updates plugin, in seconds. The minimum value is 10. To disable it, set it to -1.

Defines the path to store data in a different path than the program files directory:

• %AppDir%/data: Used for storing the delta data

• %AppDir%/user_data: External directory for user-generated JSON files

• %AppDir%/newrelic-infra.log: If log file config option is not defined, then we use this directory path as default.

  YML option name	Environment variable	Type	Default	Version
  app_data_dir	NRIA_APP_DATA_DIR	string	Windows: env(ProgramData)\New Relic\newrelic-infra Linux: Not applicable	1.0.755

Enables the Windows updates plugin, which retrieves the lists of hotfixes that are installed on the host.

If true, the agent will be forced to use Windows WMI (the agent's legacy method to grab metrics for Windows; for example, StorageSampler) and will disable the new method (which uses the PDH library).

This configuration option allows to increase the newrelic-infra.exe process priority to any of the following values:

• Normal

• Idle

• High

• RealTime

• BelowNormal

• AboveNormal

  YML option name	Environment variable	Type	Default	Version
  win_process_priority_class	NRIA_WIN_PROCESS_PRIORITY_CLASS	string		1.0.989

  Example:

  Normal

  Copy

Enables the Windows agent to report drives A: and B: when they are mapped as removable drives.

If you set it to true, you'll get process information from WMI and you'll skip the query access check on a restricted environment.

Custom HTTP headers to include in all requests to the New Relic backend.

http:
  headers:
    "Proxy-Authorization": "myToken"