Observing Python applications with OpenTelemetry
summary
Compares two zero-code paths: the OpenTelemetry Operator injects the Python agent by mounting a copied distribution and setting env vars but ships a single Python-version operator image so binary wheel/C library mismatches force building a custom image; the opentelemetry-instrument wrapper uses the agent plus ed_bootstrap auto-installs instrumentations and is runnable as a wrapped entrypoint; a third option is two-line programmatic setup that still reads env vars; GenAI work includes semantic conventions for chat/embedding spans, token and duration metrics, and choice/tool-call events.
AI-written summary · about
chapters (6)
- 0:00OpenTelemetry intro
- 5:02Kubernetes operator for Python
- 14:23Python auto-instrumentation
- 17:14Two-line instrumentation
- 18:42Observing generative AI apps
- 25:49Summary and future
related talks
30mObserving Python applications with OpenTelemetry
17mEfficient Python Observability with OpenTelemetry
24m"Introduction to OpenTelemetry with Python"
29mModern observability for Python
30mPractical introduction to OpenTelemetry Tracing
27mDistributed Tracing with OpenTelemetry
02h 02mHow To Troubleshoot and Monitor Applications using OpenTelemetry
26mWelcome to the OTEL California