random zinc
I have been trying to debug this seemingly duplicate Open telemetry tracer that appears in datadog. Seems like i have a GET GET with two returns.
Been trying to narrow it down but still stuck as before. Does Litestar have it enabled by default or something? You can see i plug my otel into the middleware.
config
"""Initialize Open Telemetry configuration."""
resource = Resource(attributes={SERVICE_NAME: settings.log.DD_SERVICE})
tracer_provider = TracerProvider(resource=resource)
processor = BatchSpanProcessor(
OTLPSpanExporter(endpoint=f"http://{settings.log.DD_AGENT_HOST}:{settings.log.DD_TRACE_AGENT_PORT}/v1/traces")
)
tracer_provider.add_span_processor(processor)
reader = PeriodicExportingMetricReader(
OTLPMetricExporter(
endpoint=f"http://{settings.log.DD_AGENT_HOST}:{settings.log.DD_TRACE_AGENT_PORT}/v1/metrics"
)
)
meter_provider = MeterProvider(resource=resource, metric_readers=[reader])
return OpenTelemetryConfig(
meter=metrics.get_meter(__name__),
middleware_class=OpenTelemetrySingletonMiddleware,
meter_provider=meter_provider,
tracer_provider=tracer_provider,
)```
normal heath
Is there any data in those 2 spans you can share?
random zinc
First span: {"asgi":{"event":{"type":"http.response.start"}},"duration":79890,"env":"none","http":{"status_code":"200"},"otel":{"library":{"name":"opentelemetry.instrumentation.asgi","version":"0.57b0"},"status_code":"Unset","trace_id":"9709ef4844038e029c1ed815d93b5cec"},"span":{"kind":"internal"}}
Second Span: {"asgi":{"event":{"type":"http.response.body"}},"duration":37107,"env":"none","otel":{"library":{"name":"opentelemetry.instrumentation.asgi","version":"0.57b0"},"status_code":"Unset","trace_id":"9709ef4844038e029c1ed815d93b5cec"},"span":{"kind":"internal"}}
normal heath
This looks correct to me though, right? The first span is the start of the response, the next is when the body started
random zinc
really? i figured the GET GET is a bug indicating some duplicate trace start
i figured the start and response so only 2 spans
normal heath
Hmm, do your other endpoints not look like this?
random zinc
they all have GET GET or POST POST. been trying to integrate datadog with open telemetry. I suspect it had something with some broken config. Because thats just one of the issues i have. when i integrate sqlalchemy, i get seemingly "orphan" spans
you can see the neondb has 1 span indicating orphan, whereas here it shows 8 span so its part of a proper trace
the waterfull looks good but the duplicate responses and the random orphan spans
wild plank