Exporters

Exporters send processed telemetry to backends. Every pipeline must end with at least one exporter.

Choosing a protocol

Use OTLP/gRPC for Collector-to-backend communication. gRPC provides better throughput and supports bidirectional streaming, which matters at the Collector's aggregation volume. Fall back to OTLP/HTTP only when network proxies do not support HTTP/2.

OTLP/gRPC exporter to Dash0

Use the OTLP/gRPC exporter to send traces, metrics, and logs to Dash0.

Where to get configuration values

1. OTLP Endpoint: In Dash0: Settings → Organization → Endpoints
2. Auth Token: In Dash0: Settings → Auth Tokens → Create Token

Minimal configuration

exporters:
  otlp:
    endpoint: <OTLP_ENDPOINT>
    headers:
      Authorization: "Bearer <AUTH_TOKEN>"

Replace <OTLP_ENDPOINT> with your Dash0 OTLP endpoint (e.g., ingress.eu-west-1.aws.dash0.com:4317). Replace <AUTH_TOKEN> with your Dash0 auth token; see the Authentication section for how to optimally set up the authentication token.

Production configuration

Configure retry, timeout, compression, and sending queue for reliable delivery.

exporters:
  otlp:
    endpoint: <OTLP_ENDPOINT>
    headers:
      Authorization: "Bearer <AUTH_TOKEN>"
    compression: gzip
    timeout: 30s
    retry_on_failure:
      enabled: true
      initial_interval: 5s
      max_interval: 30s
      max_elapsed_time: 300s
    sending_queue:
      enabled: true
      num_consumers: 10
      queue_size: 5000
      storage: file_storage

Compression

Enable gzip compression to reduce network bandwidth. The Dash0 ingress endpoint supports gzip-compressed OTLP/gRPC.

# GOOD — reduces bandwidth by 60-80 percent
exporters:
  otlp:
    endpoint: <OTLP_ENDPOINT>
    headers:
      Authorization: "Bearer <AUTH_TOKEN>"
    compression: gzip

# BAD — uncompressed traffic wastes bandwidth
exporters:
  otlp:
    endpoint: <OTLP_ENDPOINT>
    headers:
      Authorization: "Bearer <AUTH_TOKEN>"
    compression: none

Retry on failure

Enable retries to handle transient network errors and backend unavailability.

Sending queue

The sending queue buffers telemetry when the backend is temporarily unavailable. Without it, data is dropped during transient failures.

exporters:
  otlp:
    endpoint: <OTLP_ENDPOINT>
    headers:
      Authorization: "Bearer <AUTH_TOKEN>"
    sending_queue:
      enabled: true
      num_consumers: 10
      queue_size: 5000
      storage: file_storage

Use file_storage with the file storage extension to persist the queue to disk. In-memory queues lose buffered data when the Collector restarts.

Authentication

Do not hardcode auth tokens in configuration files. Reference an environment variable instead.

Create a dedicated auth token with Ingesting permissions only. The Collector needs to send telemetry, not query or manage the organization. In Dash0, create the token at Settings → Auth Tokens → Create Token and select the Ingesting scope. See auth tokens for details on available permission scopes.

# GOOD — token from environment variable
exporters:
  otlp:
    endpoint: <OTLP_ENDPOINT>
    headers:
      Authorization: "Bearer ${env:DASH0_AUTH_TOKEN}"

# BAD — hardcoded token in config
exporters:
  otlp:
    endpoint: <OTLP_ENDPOINT>
    headers:
      Authorization: "Bearer dh0_1a2b3c4d5e6f..."

Set the environment variable in your deployment manifest:

env:
  - name: DASH0_AUTH_TOKEN
    valueFrom:
      secretKeyRef:
        name: dash0-credentials
        key: auth-token

OTLP/HTTP exporter

Use the OTLP/HTTP exporter when gRPC is not available (e.g., network proxies that do not support HTTP/2).

exporters:
  otlphttp:
    endpoint: https://<OTLP_ENDPOINT>
    headers:
      Authorization: "Bearer ${env:DASH0_AUTH_TOKEN}"
    compression: gzip

The OTLP/HTTP exporter uses port 4318 by default. Check your Dash0 endpoint documentation for the correct URL.

Debug exporter

Use the debug exporter during development to print telemetry to the Collector's stdout. Do not enable the debug exporter in production — it generates excessive log output.

# GOOD — development only
exporters:
  debug:
    verbosity: detailed

# BAD — debug exporter in production pipeline
exporters:
  debug:
    verbosity: detailed
  otlp:
    endpoint: <OTLP_ENDPOINT>
    headers:
      Authorization: "Bearer ${env:DASH0_AUTH_TOKEN}"

service:
  pipelines:
    traces:
      receivers: [otlp]
      processors: [memory_limiter]
      exporters: [otlp, debug]  # debug wastes CPU and I/O in production

Verbosity levels

Multiple exporters

Send telemetry to multiple backends by listing multiple exporters in a pipeline. Each exporter receives a copy of the data independently.

exporters:
  otlp/dash0:
    endpoint: <OTLP_ENDPOINT>
    headers:
      Authorization: "Bearer ${env:DASH0_AUTH_TOKEN}"
  otlp/secondary:
    endpoint: secondary-backend:4317
    tls:
      insecure: true

service:
  pipelines:
    traces:
      receivers: [otlp]
      processors: [memory_limiter]
      exporters: [otlp/dash0, otlp/secondary]

Use named instances (otlp/dash0, otlp/secondary) to configure multiple exporters of the same type.

Anti-patterns

• Exporting without TLS. OTLP/gRPC uses TLS by default. Setting tls.insecure: true sends telemetry (including potentially sensitive data) in plaintext. Only disable TLS for local development or within a trusted network with mTLS (e.g., a service mesh).
• Hardcoding auth tokens in configuration files. Tokens in config files end up in version control, container images, and logs. Use environment variables with ${env:VARIABLE_NAME} syntax.
• Missing sending_queue in production. Without a queue, transient backend failures cause immediate data loss. Always enable the sending queue for production deployments.
• Using debug exporter in production. The debug exporter serializes every telemetry item to stdout, consuming CPU and disk I/O. Remove it from production pipelines.

References

• OTLP/gRPC exporter
• OTLP/HTTP exporter
• Debug exporter
• File storage extension