Is anyone storing open inference / otel Genai spans in elastic APM? I am having issues with the char limit of 1024 , due to which I have issues using the data to run online evals later. Is it not the correct fit for the use case? Or is there a way to get around this limit? I checked options online to use the collector and push to ES or something
Hi @nikhil_digde, welcome!
Quick thing to check before you go down the mapping route: that 1024 cut happens on the APM Server side, before the doc ever reaches Elasticsearch. Worth confirming send one long attribute and look at the raw document in Discover. If it’s already truncated there, no index template change will bring it back.
If that’s what you see, the way out is to skip the classic APM intake and store the spans in OTel-native format (Managed OTLP Endpoint on Cloud, or a collector with the elasticsearch exporter in otel mapping mode). Your collector idea was pointing the right way.
Which version and deployment type are you on?
Hey @Rafa_Silva ,
Thanks for the response. Yes I checked it and its truncated in es. So i went the collector route for now. Yes, i checked the apm es otel endpoint, but it dint work for traces. It worked for metrics only.