lievan
lievan
@maycmlee not yet; waiting on the next release of ddtrace, thanks!
Holding off on supporting metadata for custom evals
nevermind just gonna open a new one
Hey @buraizu, this isn't ready to merge yet due to some front-end updates needed - we'll let you know when it's ready, thanks!
Hey @kimwonj77 thanks for your patience here and the thorough reproduction script! Just wanted to let you know that I was able to reproduce your issue and I am working...
Hey @kimwonj77, we'll investigate how to make flushing large spans more performant. And you're right - `DD_TRACE_SAMPLING_RULES` should not affect LLM Obs in any way. As a temporary work-around, could...
Hey @kimwonj77, is your solution above enough to get all of the spans to show up (except in cases where flush times out)? > seems like flush locked up when...
Hey @kimwonj77, closing this since the work-around mentioned resolves your primary issue. I was also not able to reproduce any deadlocks, but the team will investigate a better solution for...
Hi @Gekko0114, if I understand correctly, this is a case where you 1) enable LLM Observability but then 2) want to disable data being sent in certain circumstances? If `DD_LLMOBS_ENABLED`...
Hey @Gekko0114, that is not expected behavior—LLM Observability data should not be sent if `DD_LLMOBS_ENABLED=0` is set and LLMObs is not already enabled. Unfortunately the PR does not correctly reproduce...