You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the solution you'd like
Failures to trace should have the option to do so silently. Adding an option to the TracingInterceptor to ignore the invalid SpanContext would allow us to keep this long running workflow alive when the new version of the worker deploys. It's acceptable for spans to not connect to their parents for a short time; it is not acceptable for a critical workflow to fail because of a failure to trace.
Describe alternatives you've considered
The DataDog interceptor seems to handle this gracefully already. OpenTelemetry does not return any error when getting a SpanContext.
Additional context
I've added a commit on my fork that attempts to address this. I feel like this is likely a common thing that people might run into; running workflows that straddle two tracing implementations.
The text was updated successfully, but these errors were encountered:
Maybe I'm totally off base here in thinking that workflows/activities should not function if tracing is not functioning, but I'm happy to hear any other thoughts.
Is your feature request related to a problem? Please describe.
We recently moved from OpenTracing to OpenTelemetry. Upon deploying, we found a long running workflow failed because it was trying to start a span from an OpenTracing parent span, so an error was returned here: https://github.com/temporalio/sdk-go/blob/master/contrib/opentelemetry/tracing_interceptor.go#L137
Describe the solution you'd like
Failures to trace should have the option to do so silently. Adding an option to the TracingInterceptor to ignore the invalid SpanContext would allow us to keep this long running workflow alive when the new version of the worker deploys. It's acceptable for spans to not connect to their parents for a short time; it is not acceptable for a critical workflow to fail because of a failure to trace.
Describe alternatives you've considered
The DataDog interceptor seems to handle this gracefully already. OpenTelemetry does not return any error when getting a SpanContext.
Additional context
I've added a commit on my fork that attempts to address this. I feel like this is likely a common thing that people might run into; running workflows that straddle two tracing implementations.
The text was updated successfully, but these errors were encountered: