Hi, When the OTEL collector is stopped due to some reason, the APIs which are emitting the traces to the collector start throwing the connection error. This error will not disappear even after restarting the collector. This requires the APIs to be restarted. This becomes a limitation. What is the solution for this ?

@Prashant Shahi You have any insights on this?

This could be related to instrumentation SDK. I had tried similar scenario but with HotROD sample application, I can confirm that we don't have this problem.

@manohar mirle that shouldn't be the case. the client exporter retries with exponential backoff for some time and then gives up. If collector becomes available before giving up then the traces will be exporter successfully. Which SDK are you using?

@Srikanth Chekuri yes, client exporter is retrying for some time and giving up. Is there any way to increase this retry count ? We are instrumenting Java and Python services.

@manohar mirle That's not customisable at this point. Here is an issue I created to streamlining this behaviour on spec https://github.com/open-telemetry/opentelemetry-specification/issues/1528.

Thanks @Srikanth Chekuri

Here is the policy for java https://github.com/open-telemetry/opentelemetry-java/tree/main/sdk-extensions/autoconfigure#otlp-exporter-retry and it's 64 seconds for python

@Srikanth Chekuri Is it the same policy for GOLang, FastAPI and NodeJS ?

It is framework independent

ok