Adrian Cole

I started looking into this. The timeout occurs due to saturation of ML tasks in elasticsearch. It may be possible to rewrite the initialization so that it blocks until they...

https://github.com/elastic/elasticsearch-labs/pull/397 should fix the user experience, though the time until ready could be better. In any case, this was raised about the user experience so it should close it.

PS we've reproduced that the worst of timeouts will happen on non x86 machines, which sadly means anyone on a recent macbook.

took [your demo](https://github.com/ThomasVitale/llm-apps-java-spring-ai/tree/main/10-observability/observability-models-openai) for a spin. ping back when you've span events as I'll be interested to take a second look.

@piotrooo https://github.com/ymtdzzz/otel-tui otherwise known as "missing 1000 github stars". It is technically an otel collector extension, bundled for convenience into one binary.

FWIW it feels possible for someone to add a `ObservationHandler` and make your own instrumentation which follows a convention of choosing, whether that is otel's experimental semconv, openinference, or something...

I like this as it is externalizing a hard coded default. This currently impacts users of tools like smolagents where they local test with ollama and use possibly something else...

> What would the behavior of OLLAMA_CONTEXT_LENGTH=0 be? In llama.cpp, that defaults it to the trained context length, would that be the case here? Regardless of the choice, it'd be...

ok I looked into it. I misunderstood the implementation at first. Looks like we are wrapping. I think it is a coin toss what is intuitive. Personally, I think it...

If we move to a trace processor. Wouldn't it break the ability to propagate context (since it is like a batch exporter?) Meaning you couldn't combine this with other logfire...