Viewing All Traces Within a Session: Options and Solutions

Legend H. · 2025-06-27T03:45:18.792Z

Hi, is there any way to view all traces within a session? Currently in the UI can only see 50 traces in a session.

4 comments

· Sorted by Oldest

RunLLM
APP
·
Yes, Phoenix's Sessions UI displays a paginated list of traces within a session, but by default only shows a limited number (e.g., 50) at a time. You can view all traces in a session by navigating through the pagination controls in the session details view. There is no documented way to increase the per-page limit beyond the UI's default, but all traces are accessible via pagination. For programmatic access, you can use the Python client to retrieve all spans for a session and analyze them as a dataframe using `session.get_spans_dataframe()` after filtering by session ID if needed. If you need to export or analyze all traces at once, use the Python API to pull the full set of traces for the session, as the UI is currently limited by pagination and does not support changing the page size directly. Would you like more detail or a code example for programmatic access? Sources: - <https://arize-ai.slack.com/archives/C04R3GXC8HK/p1724789781.231059|https://arize-ai.slack.com/archives/C04R3GXC8HK/p1724789781.231059> - <https://arize-ai.slack.com/archives/C04R3GXC8HK/p1726632533.889129|https://arize-ai.slack.com/archives/C04R3GXC8HK/p1726632533.889129>
💡 Hint: Mention RunLLM in the thread for followups.
Legend H.
·
RunLLM Yes please give code example
RunLLM
APP
·
To programmatically retrieve all traces within a session, use the Phoenix Python client and filter spans by the session ID. Here’s an example: ``` import phoenix as px from phoenix.trace.dsl import SpanQuery # Replace with your actual session ID session_id = "your-session-id" # Create a query to filter spans by session.id query = SpanQuery().where(f"session.id == '{session_id}'") # Retrieve all spans for the session as a DataFrame client = px.Client() spans_df = client.query_spans(query) print(spans_df) ``` This will return a DataFrame containing all traces (spans) associated with the specified session ID, regardless of UI pagination limits. You can then analyze or export this data as needed. Sources: - <https://arize.com/docs/phoenix/tracing/how-to-tracing/extract-data-from-spans#%3AS1%3A|Extract Data from Spans> - <https://arize.com/docs/phoenix/tracing/how-to-tracing/importing-and-exporting-traces/extract-data-from-spans#%3AS1%3A|Phoenix Python Client API>
💡 Hint: Mention RunLLM in the thread for followups.
Mikyo
·
Hey Legend H. probably just an oversight on our front. File us an issue and we will get it fixed!

RunLLM
APP
·
Yes, Phoenix's Sessions UI displays a paginated list of traces within a session, but by default only shows a limited number (e.g., 50) at a time. You can view all traces in a session by navigating through the pagination controls in the session details view. There is no documented way to increase the per-page limit beyond the UI's default, but all traces are accessible via pagination. For programmatic access, you can use the Python client to retrieve all spans for a session and analyze them as a dataframe using `session.get_spans_dataframe()` after filtering by session ID if needed. If you need to export or analyze all traces at once, use the Python API to pull the full set of traces for the session, as the UI is currently limited by pagination and does not support changing the page size directly. Would you like more detail or a code example for programmatic access? Sources: - <https://arize-ai.slack.com/archives/C04R3GXC8HK/p1724789781.231059|https://arize-ai.slack.com/archives/C04R3GXC8HK/p1724789781.231059> - <https://arize-ai.slack.com/archives/C04R3GXC8HK/p1726632533.889129|https://arize-ai.slack.com/archives/C04R3GXC8HK/p1726632533.889129>
💡 Hint: Mention RunLLM in the thread for followups.
Legend H.
·
RunLLM Yes please give code example
RunLLM
APP
·
To programmatically retrieve all traces within a session, use the Phoenix Python client and filter spans by the session ID. Here’s an example: ``` import phoenix as px from phoenix.trace.dsl import SpanQuery # Replace with your actual session ID session_id = "your-session-id" # Create a query to filter spans by session.id query = SpanQuery().where(f"session.id == '{session_id}'") # Retrieve all spans for the session as a DataFrame client = px.Client() spans_df = client.query_spans(query) print(spans_df) ``` This will return a DataFrame containing all traces (spans) associated with the specified session ID, regardless of UI pagination limits. You can then analyze or export this data as needed. Sources: - <https://arize.com/docs/phoenix/tracing/how-to-tracing/extract-data-from-spans#%3AS1%3A|Extract Data from Spans> - <https://arize.com/docs/phoenix/tracing/how-to-tracing/importing-and-exporting-traces/extract-data-from-spans#%3AS1%3A|Phoenix Python Client API>
💡 Hint: Mention RunLLM in the thread for followups.
Mikyo
·
Hey Legend H. probably just an oversight on our front. File us an issue and we will get it fixed!