Hey everyone! We're trying to instrument a JS app using a combination of langchain + openai autoinstrumentors with custom instrumentation (basically using custom instrumentation to capture all spans from langchain/openai autoinstrumentors under the same trace), and we're seeing some duplicate spans popping up from the manual instrumentation in Phoenix. It looks like the duplicate spans are duplicates of the root span only. We're using the BatchSpanProcessor but when we switch to SimpleSpanProcessor the duplicate spans go away. Does anyone know why the batch span processor duplicates some spans with some of the duplicates being orphaned(no parent id)?
Not sure Trevor L. - typically we don't see a need to instrument both of these together as the langchain instrumentation is pretty comprehensive. We have a working example could you maybe reproduce this issue and file a bug for us if you can reproduce it? Sorry we are pretty busy so a minimum working example would be great. In general duplicate spans are not really possible in Phoenix so it's pretty puzzing. https://github.com/Arize-ai/openinference/tree/main/js/examples/langchain-express
