Request for Granular Time Filters on Traces View

Ben M. · 2024-12-18T21:26:33.053Z

Is it possible to set more granular time filters on the traces view? Rather than 1 day or 7 days, I would love to choose a start and end time

6 comments

· Sorted by Oldest

RunLLM
APP
·
Yes, it is possible to set more granular time filters on the traces view in Phoenix. You can specify a time range using the start_time and end_time parameters when querying spans. This allows you to focus on specific time frames, such as hours or minutes, rather than being limited to broader ranges like 1 day or 7 days. This feature is particularly useful for troubleshooting or analyzing specific periods of data ingestion.
Would you like more detail on how to implement this?
Sources: - Extract Data from Spans - Slack Discussion on Time Range
💡 Hint: Mention RunLLM in the thread for followups.
Ben M.
·
RunLLM what is a query that I could use when looking at traces? I don’t see that in the autocomplete
RunLLM
APP
·
To query traces in Phoenix, you can use the query_spans method with a SpanQuery object. This allows you to filter spans based on specific criteria. For example, you can filter spans by span_kind or other attributes. Here's a basic example of how to query for spans of a specific kind:
```
import phoenix as px from phoenix.trace.dsl import SpanQuery
query = SpanQuery().where("span_kind == 'RETRIEVER'")
df = px.Client().query_spans(query)
LBKSWould you like more detail on constructing queries or using other filters? Sources: - [Extract Data from Spans](https://docs.arize.com/phoenix/tracing/how-to-tracing/extract-data-from-spans#tS7q1citZXMJ)
💡 Hint: Mention RunLLM in the thread for followups.
Ben M.
·
I was able to scroll back, but had performance issues due to the amount and size of spans in our 7 day window. The data was just over 24 hours old, so I couldn’t use the 1 day filter. It would be nice to input a time window dynamically like you see in a lot of logging platforms like AWS Cloudwatch
Mikyo
·
Hey Ben M. this is definitely something that's super needed. 😅 It's captured here https://github.com/Arize-ai/phoenix/issues/2596 - definitely upvote and keep the feedback coming. I feel your pain. We'll try to tackle it soon.
Ben M.
·
Thank you! I upvoted the issue and will absolutely give feedback as I encounter pain points 🙂

Request for Granular Time Filters on Traces View | Arize AI Community

RunLLM
APP
·
Yes, it is possible to set more granular time filters on the traces view in Phoenix. You can specify a time range using the start_time and end_time parameters when querying spans. This allows you to focus on specific time frames, such as hours or minutes, rather than being limited to broader ranges like 1 day or 7 days. This feature is particularly useful for troubleshooting or analyzing specific periods of data ingestion.
Would you like more detail on how to implement this?
Sources: - Extract Data from Spans - Slack Discussion on Time Range
💡 Hint: Mention RunLLM in the thread for followups.
Ben M.
·
RunLLM what is a query that I could use when looking at traces? I don’t see that in the autocomplete
RunLLM
APP
·
To query traces in Phoenix, you can use the query_spans method with a SpanQuery object. This allows you to filter spans based on specific criteria. For example, you can filter spans by span_kind or other attributes. Here's a basic example of how to query for spans of a specific kind:
```
import phoenix as px from phoenix.trace.dsl import SpanQuery
query = SpanQuery().where("span_kind == 'RETRIEVER'")
df = px.Client().query_spans(query)
LBKSWould you like more detail on constructing queries or using other filters? Sources: - [Extract Data from Spans](https://docs.arize.com/phoenix/tracing/how-to-tracing/extract-data-from-spans#tS7q1citZXMJ)
💡 Hint: Mention RunLLM in the thread for followups.
Ben M.
·
I was able to scroll back, but had performance issues due to the amount and size of spans in our 7 day window. The data was just over 24 hours old, so I couldn’t use the 1 day filter. It would be nice to input a time window dynamically like you see in a lot of logging platforms like AWS Cloudwatch
Mikyo
·
Hey Ben M. this is definitely something that's super needed. 😅 It's captured here https://github.com/Arize-ai/phoenix/issues/2596 - definitely upvote and keep the feedback coming. I feel your pain. We'll try to tackle it soon.
Ben M.
·
Thank you! I upvoted the issue and will absolutely give feedback as I encounter pain points 🙂