Replies: 2 comments 1 reply
-
hi @lmssy! The agent and SDK use "head-based" sampling, making their decision at the start of each span, when the error status is not available yet. You may want to look into tail-sampling provided by the OpenTelemetry Collector. Or alternatively you could possibly implement some kind of tail-sampling in the SpanProcessor pipeline for your exporter depending on your needs. |
Beta Was this translation helpful? Give feedback.
-
@trask Would this answer change now in 2024? I wanted to configure sampling on the application level and have errors not sample also on the application level, instead of sampling at the collector. Is this possible? |
Beta Was this translation helpful? Give feedback.
-
Hi,
I'm using auto instrumentation to my server and set smapler rate to 0.01, I konw not all spans will be exported to collector.
But I want to make sure all spans with error (like application's exception or those traceId/spanId recorded in error log ) can be send to collector, is there any way to achieve?
Beta Was this translation helpful? Give feedback.
All reactions