modra40 OpenTelemetry Mastery (Senior → Principal)
Operate
- Start from what must be explainable in production: request flow, failure attribution, dependency latency, and causality.
- Treat telemetry schemas, propagation, sampling, and collector design as platform architecture.
- Prefer standardized semantic conventions over one-off instrumentation creativity.
- Design for correlation across traces, metrics, and logs without uncontrolled cost.
Default Standards
- Context propagation must be reliable across service and async boundaries.
- Sampling policy should match debugging and business-critical workflows.
- Collector pipelines should be simple enough to operate and debug.
- Telemetry attributes are a cost and governance budget.
- Vendor portability should not mean lowest-common-denominator blindness.
References
- Tracing model and span design: references/tracing-model-and-span-design.md
- Context propagation and async boundaries: references/context-propagation-and-async-boundaries.md
- Semantic conventions and schema governance: references/semantic-conventions-and-schema-governance.md
- Sampling strategy and cost control: references/sampling-strategy-and-cost-control.md
- Collector architecture and pipelines: references/collector-architecture-and-pipelines.md
- Metrics, logs, and correlation: references/metrics-logs-and-correlation.md
- Reliability and operations: references/reliability-and-operations.md
- Incident runbooks: references/incident-runbooks.md