Most OpenTelemetry pain is self-inflicted and predictable. Here are the mistakes we see most, and the fix for each.
Mistake 1: Sampling misconfigured, so you keep everything and blow up cost
Fix: Send everything to a Collector, then fan out to backends. Make it the default in your templates and enforce it in review or CI so the mistake cannot come back.
Mistake 2: Mixing SDK versions and losing context propagation
Fix: Use tail-based sampling for traces to keep the interesting ones. Make it the default in your templates and enforce it in review or CI so the mistake cannot come back.
Mistake 3: Exporting directly from apps instead of through a Collector
Fix: Propagate context across services with W3C tracecontext. Make it the default in your templates and enforce it in review or CI so the mistake cannot come back.
A safer baseline
A Collector pipeline: receive OTLP, batch, export.
receivers:
otlp: { protocols: { grpc: {}, http: {} } }
processors:
batch: {}
exporters:
prometheus: { endpoint: 0.0.0.0:8889 }
otlphttp: { endpoint: https://tempo:4318 }
service:
pipelines:
traces: { receivers: [otlp], processors: [batch], exporters: [otlphttp] }
metrics: { receivers: [otlp], processors: [batch], exporters: [prometheus] }Want senior help putting OpenTelemetry to work - or a team bootcamp on it? CodeWizard runs Israel's largest DevOps community and embeds named senior engineers with a transparent hours log. Book a free consult or explore the Academy.