I'm struggling to interpret this output from a trace of a java service for kafka.consume in Datadog. I would expect the parent (kafka.consume) to span the entire duration of the child spans:
You can see the kafka.consume is tracked as 14.9 s duration, but the child spans exceed that. What is going on?
I don't know how your program is written, but
kafka.consume
is likely running in its own thread. This trace indicates thatkafka.consume
triggeredMemberCosmosServiceImpl.getMe...
asynchronously and did not wait forMemberCosmosService...
to return something.