Python SDK: How to Retrieve Most Recent Spans Data

Suhrullekha C. · 2025-08-07T13:07:08.670Z

hey guys, I'm trying to export spans data using the python sdk (get_spans_dataframe) and I'm getting a response which is 4 hours behind my most rencent run. I haven't added any filter and made sure to change the timezone. Is there a way to get the most recent data?

RunLLM
APP
·
By default, Phoenix span queries use UTC for filtering, regardless of your local timezone. If you want to get the most recent data, explicitly set the start_time and end_time parameters in your query using UTC-aware datetimes. This ensures you retrieve the latest spans, not just those matching your local time offset. Example: LBKSfrom datetime import datetime, timedelta, timezone import phoenix as pxLBKSpx_client = px.Client() start = datetime.now(timezone.utc) - timedelta(hours=1) # last 1 hour in UTC spans_df = px_client.query_spans(start_time=start)LBKSAccording to the documentation, all time-based filters are applied in UTC, not your local timezone, so local adjustments alone won't affect the query results. Always use UTC when specifying time ranges for accurate, up-to-date data retrieval (source, source, source).Would you like more detail or troubleshooting steps?Sources: - https://github.com/Arize-ai/phoenix/issues/3024 - https://arize.com/docs/phoenix/tracing/how-to-tracing/importing-and-exporting-traces/extract-data-from-spans#%3AS1%3A - https://arize.com/docs/phoenix/how-to/extract-data-from-spans
💡 Hint: Mention RunLLM in the thread for followups.
Suhrullekha C.
·
hey 🔒[private user] when i use query_spans instead and specify the time range i dont get anything back. Can i pass start_time and end_time as parameters while using the get_spans_dataframe()?
Suhrullekha C.
·
Phoenix cloud.
Suhrullekha C.
·
Suhrullekha C.
·
sounds good, thanks. And yes they are since I'm getting a response when i use get_spans_dataframe
:thxxx:1
Xander S.
·
Suhrullekha C. What time zone are you in when running the Phoenix client + what time zone is your application emitting traces from?
Xander S.
·
Can you also show me the code you use to invoke get_spans_dataframe?
Roger Y.
·
can you try adding timezone to datetime.now like this (from datetime import timezone)
datetime.now(timezone.utc)
Suhrullekha C.
·
Xander S. I'm running it in "America/Toronto" timezone and this is how I set up the get_spans_dataframe
Suhrullekha C.
·
Suhrullekha C.
·
Roger Y. I'm still getting no response when i update the timezone and send it in to query_spans()
Xander S.
·
Suhrullekha C. What do your timestamps look like in the dataframe before you convert the timezone? Are they timezone-aware or timezone-naive?
Suhrullekha C.
·
Xander S. the time just shows up as this 2025-08-06 17:41:39.852124+00:00 which indicates its utc
👍1
Roger Y.
·
just to clarify: when start and end time are not specified you’re getting data but the data is to old, is that correct?
Roger Y.
·
but when start and end times are specified, no data is returned. is that correct?

25 comments