OTLP Profiles Ingestion

Send profiles to OpenObserve over OTLP, the same protocol used for logs, metrics, and traces.

Authorization

HTTP and gRPC both send Authorization: Basic <base64(email:password)>. Generate the value with:

printf '%s' 'your-email:your-password' | base64 | tr -d '\n'

You can also copy the endpoint and header from Data Sources → Custom → Profiles.

OTLP/HTTP

The resource processor in these examples assigns a name to a single application. For a Collector receiving multiple services, omit this processor and have each producer supply service.name; upserting one constant name groups them together in the UI.

Send an ExportProfilesServiceRequest with Content-Type: application/x-protobuf or application/json. Raw pprof and JFR files are not OTLP; convert them with the integrations linked below.

receivers:
  otlp:
    protocols:
      http:
        endpoint: 0.0.0.0:4318

processors:
  resource:
    attributes:
      - key: service.name
        value: your-service
        action: upsert

exporters:
  otlp_http/openobserve:
    profiles_endpoint: http://<your-openobserve-host>/api/<your-org>/v1/profiles
    headers:
      Authorization: Basic <paste-from-Data-Sources>
      stream-name: default

service:
  pipelines:
    profiles:
      receivers: [otlp]
      processors: [resource]
      exporters: [otlp_http/openobserve]

otelcol-contrib --feature-gates=+service.profilesSupport --config=otelcol-config.yaml

OTLP/gRPC

Default port is 5081 (ZO_GRPC_PORT). The organization travels in gRPC metadata, not in the endpoint path. The example uses plaintext for a local server (tls.insecure: true); use insecure: false and the appropriate CA configuration for a TLS endpoint.

receivers:
  otlp:
    protocols:
      grpc:
        endpoint: 0.0.0.0:4317

processors:
  resource:
    attributes:
      - key: service.name
        value: your-service
        action: upsert

exporters:
  otlp_grpc/openobserve:
    endpoint: <your-openobserve-host>:5081
    headers:
      Authorization: "Basic <paste-from-Data-Sources>"
      organization: <your-org>
      stream-name: default
    tls:
      insecure: true

service:
  pipelines:
    profiles:
      receivers: [otlp]
      processors: [resource]
      exporters: [otlp_grpc/openobserve]

otelcol-contrib --feature-gates=+service.profilesSupport --config=otelcol-config.yaml

Verify ingestion

1. Start the Collector with the Profiles feature gate and send a profile from one of the integrations below.
2. Check the Collector logs for rejected or dropped exports. An HTTP 200 from the Collector receiver does not confirm that OpenObserve accepted the downstream export; also check any OTLP partial_success response.
3. In OpenObserve, open Profiles, select the destination stream (default unless you changed stream-name), and select the service, profile type, and unit. Use Last 15 minutes for a fresh capture.
4. Confirm that the time series and Flame Graph, Call Tree, or Top view contain samples. Profile type labels and units come from the producer; select the emitted pair rather than assuming every CPU profile is named cpu.

If a Collector 0.153.0 pprof export is rejected with dictionary.mapping_table[0] must be the zero value, OpenObserve is rejecting a compacted dictionary with data at index zero. Upgrade OpenObserve to a version that accepts compacted Collector dictionaries; changing the endpoint or credentials does not fix that error.

Next steps

• Java / async-profiler
• Go / pprof
• Rust / pprof
• eBPF profiler
• Profiles overview

Need some help?

• Join our Community Slack
• Or Contact support