Seth D.

(yeah, it's fine by me)

Where do I sign? 🙂

If you could inquire about that being acceptable, I'd appreciate it - but I will also try and identify some modestly-sized queries that will let me export little slices of projects/dates of interest - thanks, Roger (et al)

Sure, totally understand that there are operational considerations that may prevent that from being allowable

Actually, that would be great, thanks - as long as there's nothing proprietary/secret to you guys in there, of course

Yeah, I don't imagine it'll want to slurp down that all as one big parquet file

(Not sure how/where that ever worked - maybe does when all the app is doing is exporting traces to you?)

🔒[private user] - my apologies - it would appear that working in the same environment as some of the application code that we record traces from has introduced an different value for PHOENIX_COLLECTOR_ENDPOINT - was using https://app.phoenix.arize.com/v1/tracesinstead of https://app.phoenix.arize.com

Don't also need to be supplying a OTEL_EXPORTER_OTLP_HEADERS do I?

Sure - as regards headers, it's via an environment variable (PHOENIX_CLIENT_HEADERS with a value of api_key=<PRIVATE_CRUD_HERE>) The call attempted that's indicating an underlying HTTP call got a 403 is just what's shown @ https://arize.com/docs/phoenix/tracing/how-to-tracing/importing-and-exporting-traces/extract-data-from-spans#save-all-traces

some_trace_id = px.Client().get_trace_dataset().save(directory=directory)