Custom OTel Collector

Adding a CustomOTelCollector to your namespace gives your application a dedicated OpenTelemetry Collector and automatic instrumentation — with no cluster-admin access required.

What it provides

When you create a CustomOTelCollector in your namespace, two resources are created alongside it:

Resource	Name	Purpose
OpenTelemetryCollector	<your-name>	Receives, processes, and exports telemetry from your pods
Instrumentation	grpc	Auto-instruments Java, .NET, and Go workloads via a pod annotation

The collector is pre-configured with exporters for traces (Tempo), logs (Loki), and optionally Splunk. You only configure what is specific to your application.

When to use it

Use a CustomOTelCollector when:

• Your application needs a dedicated OTel pipeline (e.g. custom log filters).
• You want auto-instrumentation for Java, .NET, or Go without writing OTel SDK code.
• You need to route logs to Splunk in addition to the default flow pipeline.

If your application has no special pipeline requirements, ask your TABBer whether a shared collector already exists in the namespace.

Quick start

Add a CustomOTelCollector to your Helm chart and annotate your Deployment.

1. Add the resource to your chart

Create templates/otel.yaml in your Helm chart:

custom-otel-collector-example.yaml

apiVersion: k8s.flow.rws.nl/v1alpha1
kind: CustomOTelCollector
metadata:
  name: {{ include "my-app.fullname" . }}-otel
  namespace: {{ .Release.Namespace }}
spec: {}

That’s the minimal configuration. Crossplane creates the collector and instrumentation automatically.

2. Enable auto-instrumentation on your Deployment

You need to add an annotation to the spec.template.metadata of your Deployment. The annotation name must match the CustomOTelCollector name in the same namespace. If you named it my-app-otel, set the annotation value to my-app-otel:

instrumentation.opentelemetry.io/inject-java: "my-app-otel"

3. Optional — custom log pipelines and processors

To add a log filter and route logs to a Splunk instance:

custom-otel-collector-splunk-example.yaml

apiVersion: k8s.flow.rws.nl/v1alpha1
kind: CustomOTelCollector
metadata:
  name: {{ include "my-app.fullname" . }}-otel
  namespace: {{ .Release.Namespace }}
spec:
  extraProcessors:
    filter/audit_logs:
      error_mode: ignore
      logs:
        log_record:
          - 'not IsMatch(body, ".*AUDIT.*")'
  pipelines:
    splunk:
      extraProcessors:
        - filter/audit_logs
    flow:
      extraProcessors:
        - filter/audit_logs

4. Optional — pin auto-instrumentation images

Override the cluster-default instrumentation image per collector:

custom-otel-collector-pin-image-example.yaml

spec:
  instrumentation:
    grpc:
      java:
        image: ghcr.io/open-telemetry/opentelemetry-operator/autoinstrumentation-java:2.5.0

Spec reference

Tip

Omit spec.pipelines.splunk entirely if you do not need Splunk log export. The logs/splunk pipeline is only created when pipelines.splunk is present.

Field	Type	Description
extraProcessors	object	Named OTel processors merged into the collector. Keys are processor names, values are processor config.
pipelines.splunk.extraProcessors	[]string	Processor names added to the logs/splunk pipeline. Omit pipelines.splunk to skip this pipeline.
pipelines.flow.extraProcessors	[]string	Processor names added to the logs/flow pipeline. Omit pipelines.flow to skip this pipeline.
instrumentation.grpc.dotnet.image	string	Override .NET instrumentation image for this collector.
instrumentation.grpc.go.image	string	Override Go instrumentation image for this collector.
instrumentation.grpc.java.image	string	Override Java instrumentation image for this collector.