This page lists the Agent Control settings you can modify, with their default values. For an overview of local and remote configuration, and for examples of how to configure the agents that Agent Control manages, see Managing configurations.
The Fleet Control connection (fleet_control), host_id, the managed agents, and value providers are covered in Managing configurations.
How Agent Control reads its settings
Agent Control reads its settings from the local configuration file:
Environment | Local configuration file |
|---|---|
Linux |
|
Windows |
|
Kubernetes | The |
Important
Only the agents setting and the version settings used for upgrades can be changed through remote configuration from Fleet Control. All other settings on this page are local: change them in the local configuration file or with environment variables, and restart Agent Control to apply them.
Override settings with environment variables
You can set any setting as an environment variable. Use the NR_AC_ prefix, write the setting name in uppercase, and use a double underscore (__) to separate nested keys. Environment variables take precedence over the local configuration file.
$# log.level$NR_AC_LOG__LEVEL=DEBUG$
$# fleet_control.auth_config.client_id$NR_AC_FLEET_CONTROL__AUTH_CONFIG__CLIENT_ID="some-client-id"On hosts, you can define these variables in the service environment file or in an environment_variables.yaml file in the local configuration directory (/etc/newrelic-agent-control/ on Linux, C:\Program Files\New Relic\newrelic-agent-control\ on Windows). The file must be a flat YAML map of string keys to string values:
NR_AC_LOG__LEVEL: "DEBUG"HTTPS_PROXY: "http://proxy.local:8443"- Keys must start with a letter or underscore, and contain only letters, digits, and underscores.
- Agent Control doesn't watch this file. Restart Agent Control after you change it.
- Agent Control passes these variables to the agents it manages.
- If the file has an invalid entry, Agent Control fails to start.
Logging: log
Use the log setting to change the log level, enable file logging, or change the log format.
log: level: debug format: formatter: pretty target: true file: enabled: true path: /var/log/my-folder/agent-control.logSetting | Description | Default |
|---|---|---|
| Log level: |
|
| A tracing filter directive (for example, | (empty) |
|
|
|
| Include the log target (the module that emitted the log) in each log line. |
|
| Timestamp format, using strftime specifiers. |
|
| Use ANSI colors in the logs sent to standard error. |
|
| Write logs to a file. |
|
| Path of the log file. If you don't set it, Agent Control writes | (default log directory) |
| Show tracing spans as log lines. Spans add details that can help with debugging. |
|
Tip
Log files rotate daily, and Agent Control keeps the last 30 files. You can't change this behavior.
Proxy: proxy
Agent Control uses the HTTP_PROXY and HTTPS_PROXY environment variables of your system. You can also set a proxy with the proxy setting. If more than one is defined, Agent Control uses them in this order of precedence:
- The
proxysetting - The
HTTP_PROXYenvironment variable - The
HTTPS_PROXYenvironment variable
proxy: url: https://user:password@proxy.example.com:8080 ca_bundle_file: /etc/ssl/certs/proxy-ca.pemSetting | Description | Default |
|---|---|---|
| Proxy URL, in the format | (empty) |
| Directory with CA certificates in PEM format. Agent Control reads all | (empty) |
| Path to a CA certificate bundle in PEM format. | (empty) |
| Ignore the |
|
Caution
The proxy you configure in Agent Control isn't applied automatically to the agents it manages. Configure the proxy separately for each agent.
Status server: server
The local status server reports the status of Agent Control and its managed agents. For how to query it, see Monitoring with Agent Control.
server: enabled: true host: "127.0.0.1" port: 51200Setting | Description | Default |
|---|---|---|
| Enable the status server. |
|
| Address the status server listens on. |
|
| Port the status server listens on. |
|
Health checks: health_check
health_check: interval: 60s initial_delay: 60sSetting | Description | Default |
|---|---|---|
| Time between Agent Control health checks. |
|
| Time to wait before the first health check. |
|
Uptime report: uptime_report
Agent Control reports its uptime as a metric when self-instrumentation metrics are enabled.
uptime_report: enabled: true interval: 60sSetting | Description | Default |
|---|---|---|
| Report the uptime of Agent Control. |
|
| Time between uptime reports. |
|
Self-instrumentation: self_instrumentation
Agent Control can report its own logs and metrics using OpenTelemetry. If you configure a proxy, self-instrumentation uses it.
self_instrumentation: opentelemetry: endpoint: https://otlp.nr-data.net:4318 headers: api-key: YOUR_LICENSE_KEY metrics: enabled: true interval: 120s logs: enabled: trueSetting | Description | Default |
|---|---|---|
| HTTPS endpoint that receives the OTLP data. Required when | (none) |
| Headers added to every request to the endpoint. | (empty) |
| Timeout for requests to the endpoint. |
|
| Attributes added to all reported metrics and logs. | (empty) |
| Filter directive that defines which logs are reported. It's insecure because enabling external dependencies can leak sensitive data. The default only includes Agent Control components. |
|
| Report metrics. |
|
| Time between metric reports. |
|
| Report logs. |
|
| Delay between batch exports of logs. |
|
| Maximum number of logs in a single batch. |
|
OCI registry: oci
Agent Control pulls agent packages from an OCI registry.
oci: registry: "docker.io" auth: basic: username: "my-user" password: "my-password"Setting | Description | Default |
|---|---|---|
| Registry host name or URL. It can't be empty. |
|
| Credentials for basic authentication. | Anonymous access |
| Token for bearer authentication. | Anonymous access |
If you set oci.auth, define exactly one of basic or bearer.
Package and agent type verification
Agent Control verifies the signatures of the packages it downloads and of the agent type definitions it pulls. Only disable verification for testing: Agent Control logs a warning when it's disabled.
agent_packages: signature_verification_enabled: true
agent_types: default_remote: signature_verification_enabled: trueSetting | Description | Default |
|---|---|---|
| Verify the signature of agent packages. |
|
| Verify the signature of agent type definitions. |
|
| Repository, in the configured | New Relic's agent types repository |
| JWKS URL of the public key used to verify agent type signatures. | New Relic's public key URL |
Self-update on hosts: self_update
On hosts, Agent Control can update itself when Fleet Control deploys a new version.
self_update: download_retry: max_attempts: 3 upgrade_backoff: base_delay: 30s max_delay: 600sSetting | Description | Default |
|---|---|---|
| Verify the signature of the Agent Control package. |
|
| OCI repository that holds the Agent Control packages. | New Relic's repository |
| JWKS URL of the public key used to verify the package signature. | New Relic's public key URL |
| Maximum download attempts per update attempt. Must be greater than zero. |
|
| Initial delay between download retries. |
|
| Maximum delay between download retries. |
|
| Add random jitter to the retry delays. |
|
| Number of consecutive failed upgrades before Agent Control escalates the failure report. It keeps retrying at |
|
| Initial wait before retrying a failed upgrade. |
|
| Maximum wait before retrying a failed upgrade. |
|
| Add random jitter to the upgrade retry delays. |
|
Kubernetes settings: k8s
The Agent Control Helm chart sets most of the k8s settings for you. Change them only if you manage the deployment yourself.
k8s: cluster_name: "my-cluster" namespace: "newrelic-agent-control" namespace_agents: "newrelic-agent-control-agents"Setting | Description | Default |
|---|---|---|
| Required. Name that identifies the cluster in Fleet Control. | (none) |
| Required. Namespace where Agent Control creates the resources it manages directly. | (none) |
| Required. Namespace where Agent Control creates the resources of the managed agents. | (none) |
| Version of the chart used to deploy Agent Control. The Helm chart passes it as an environment variable to avoid race conditions after a failed upgrade. | (empty) |
| Apply remote updates of the Agent Control chart. When |
|
| Helm release name of Agent Control. If it's empty, Agent Control doesn't check its own health or apply updates. | (empty) |
| Whether the continuous deployment (Flux) component is enabled. When |
|
| Apply remote updates of the Agent Control CD chart. When |
|
| Helm release name of the Agent Control CD chart. | (empty) |
| Name of the Kubernetes secret that holds the credentials. |
|
| Key in the secret that holds the credential value. |
|