OpenTelemetry Collector Config Generator

The OpenTelemetry Collector is a vendor-agnostic proxy that receives, processes, and exports telemetry data (traces, metrics, logs). Instead of each application sending directly to your backend (Jaeger, Prometheus, Datadog, etc.), you point them all to the Collector. This decouples your instrumentation from your backend — you can change backends without touching application code, and the Collector handles batching, retry, and enrichment centrally.

Receivers ingest telemetry data into the Collector (OTLP over gRPC/HTTP, Jaeger, Prometheus scrape, Zipkin, etc.). Processors transform or filter data in-flight — the batch processor groups spans before export, memory_limiter prevents OOM, and filter/transform processors can drop or enrich data. Exporters send processed data to your backend. You wire these together in pipelines: receivers → processors → exporters. A single Collector can run multiple independent pipelines.

The batch processor accumulates spans/metrics/logs and sends them in groups rather than individually. This dramatically reduces the number of export requests, lowering network overhead and backend load. Key settings: send_batch_size (max items per batch, default 8192), timeout (max wait before flushing, default 200ms), send_batch_max_size (hard cap on batch size). For production, always include the batch processor as the last processor before exporters.

The memory_limiter periodically checks the Collector's heap size. When usage exceeds limit_mib, it starts refusing new data and forces a GC cycle. When it exceeds spike_limit_mib above the soft limit, it drops data. This prevents the Collector from crashing and taking down your entire telemetry pipeline. Always place memory_limiter as the FIRST processor in your chain so it checks memory before any expensive transformations.

OTLP (OpenTelemetry Protocol) is the native wire format. The OTLP exporter sends to any OTLP-compatible backend: Jaeger (4317 gRPC), Tempo, Collector-to-Collector, Honeycomb, Datadog Agent, etc. For gRPC use endpoint: host:4317 (no scheme prefix). For HTTP use endpoint: http://host:4318. If your backend requires TLS, configure tls: insecure: false and provide cert_file/key_file. Most modern backends prefer OTLP over their native protocols.