OpenTelemetry

Most important commands to remember

• otelcol --version — identify the installed Collector build.
• otelcol validate --config=FILE — check a pipeline configuration without starting it.

Commands and flags

Command or syntax	Meaning
--version	Print the Collector version.
otel_config=$(mktemp)	Create a temporary configuration file and save its path.
cat > FILE <<'YAML'	Write the literal YAML block without shell expansion.
validate --config=FILE	Load and validate that configuration.
rm FILE	Remove the test configuration.

The YAML declares an OTLP/HTTP receiver on loopback port 4318, a batch processor, and a debug exporter. service.pipelines.traces connects them in that order. Validation does not open the listener or send telemetry.

The concepts that matter

1. Instrumentation creates the observations

OpenTelemetry supplies APIs, SDKs, semantic conventions, and tooling for telemetry such as traces, metrics, and logs. Application instrumentation records observations where work actually happens.

Automatic instrumentation can cover supported frameworks and libraries, while manual instrumentation can add business-specific operations. Installing a Collector alone does not magically expose every application operation. The source still needs to produce the relevant signal.

2. OTLP carries telemetry between components

The OpenTelemetry Protocol, or OTLP, transports telemetry to a Collector or compatible backend. HTTP and gRPC variants have different endpoint and protocol requirements.

A reachable port is not enough: exporter protocol, endpoint path where applicable, TLS, and authentication must match the receiver. Resource attributes such as service identity help the receiving system associate observations with the correct workload.

3. A Collector pipeline receives, processes, and exports

A receiver accepts or gathers data. Processors can batch, filter, or transform it. An exporter sends it onward. Configuration must both define components and connect them in the appropriate signal pipeline.

Processor order matters because each stage sees the previous stage’s result. Available components also depend on the Collector distribution and version. A configuration copied from a different build may name a component that is not included in yours.

4. Telemetry needs its own operational care

The Collector is generally a processing and transport layer, not a complete long-term query database or dashboard. The backend stores and makes observations searchable according to its own capabilities.

Queues, batching, retry policy, resource limits, and destination availability influence data loss and delay. Sensitive attributes can also pass through the pipeline unless deliberately handled. Observe the Collector itself so a missing trace is not automatically blamed on the instrumented application.

One small example

Optional: inspect the installed Collector version, create the temporary file, and validate it. The example does not run a receiver, require an application, or need a backend account.

otelcol --version
otel_config=$(mktemp)
cat > "$otel_config" <<'YAML'
receivers:
  otlp:
    protocols:
      http:
        endpoint: 127.0.0.1:4318
processors:
  batch: {}
exporters:
  debug: {}
service:
  pipelines:
    traces:
      receivers: [otlp]
      processors: [batch]
      exporters: [debug]
YAML
otelcol validate --config="$otel_config"
rm "$otel_config"

A successful validation means this build accepts the component configuration and pipeline structure. An unknown component or unsupported command indicates a distribution/version mismatch to resolve. It is not evidence of an application export failure.

The debug exporter would emit diagnostic telemetry if the pipeline ran; it is not durable storage. Because this command only validates, no spans should be expected in any backend. The final line removes the temporary file. Runtime connectivity, batching, and successful export remain untested.

Keep this idea: OpenTelemetry standardizes producing and moving observations; instrumentation, Collector processing, and backend storage are separate responsibilities.