Telemetry and data retention

These settings control the telemetry your self-hosted data plane sends to Braintrust’s control plane, and the service token that enables data retention.

Enable or disable telemetry

Braintrust can send the following types of telemetry from your self-hosted data plane to Braintrust’s control plane:

Type	Description
`status`	Health check information (enabled by default)
`metrics`	System metrics (CPU/memory) and Braintrust-specific metrics like indexing lag (enabled by default)
`usage`	Billing usage telemetry for aggregate usage metrics (enabled by default)
`memprof`	Memory profiling statistics and heap usage patterns
`logs`	Application logs
`traces`	Distributed tracing data

By default, status, metrics, and usage are enabled. You can change the defaults as follows:

AWS
GCP / Azure

Add the monitoring_telemetry variable to your variables.tf file, and include the types of telemetry you want to send in the validation condition as a comma-separated list:

variable "monitoring_telemetry" {
  description = <<-EOT
    The telemetry to send to Braintrust's control plane to monitor your deployment. Should be in the form of comma-separated values.

    Available options:
    - status: Health check information (default)
    - metrics: System metrics (CPU/memory) and Braintrust-specific metrics like indexing lag (default)
    - usage: Billing usage telemetry for aggregate usage metrics
    - memprof: Memory profiling statistics and heap usage patterns
    - logs: Application logs
    - traces: Distributed tracing data
  EOT
  type        = string
  default     = "status,metrics,usage"

  validation {
    condition = var.monitoring_telemetry == "" || alltrue([
      for item in split(",", var.monitoring_telemetry) :
      contains(["metrics", "logs", "traces", "status", "memprof", "usage"], trimspace(item))
    ])
    error_message = "The monitoring_telemetry value must be a comma-separated list containing only: metrics, logs, traces, status, memprof, usage."
  }
}

Update the controlPlaneTelemetry setting in your Helm values.yaml file to include the types of telemetry you want to send:

# Global configs
global:
  orgName: "<your org name on Braintrust>"
  # When createNamespace is true, the namespace will be created and resources will be in global.namespace
  # When createNamespace is false, resources will use .Release.Namespace (the namespace specified during helm install/upgrade)
  createNamespace: false
  namespace: "braintrust"
  namespaceAnnotations: {}
  labels: {}
  controlPlaneTelemetry: "status,metrics,usage,logs,traces,memprof"

Braintrust also has access to endpoints reporting metrics about the backfill and compaction status of Brainstore segments. This is metadata only, no customer data. To disable these endpoints, set the DISABLE_SYSADMIN_TELEMETRY environment variable to true.

If you disable telemetry, Braintrust’s ability to proactively monitor your deployment and diagnose issues will be significantly limited. Before disabling, consider the impact on support response times.

Data retention

Data retention requires a service token so the data plane can query object metadata and look up retention policies configured in your organization. Braintrust automatically provisions this token when you configure your self-hosted data plane URL in organization settings. It is created with read-only permissions on projects and stored securely in your data plane. To verify or refresh the token, go to Settings > Service tokens. If the token doesn’t exist, click Create. To rotate it, click Refresh — the data plane will start using the new token automatically.

Before configuring retention, review your cloud provider’s bucket retention policies. See Cloud provider retention policies for details.

Start

Instrument

Observe

Annotate

Evaluate

Deploy

Admin

Best practices

Telemetry and data retention

Enable or disable telemetry

Data retention

​Enable or disable telemetry

​Data retention

Enable or disable telemetry

Data retention