medium10 pts · first answer countstracingperformance

Reading a slow trace

A Jaeger trace for a slow 900 ms request shows the handler span calling five independent backend spans of about 170 ms each, one after another with no overlap. What is the most effective improvement?

A.Increase the sampling rate so more traces are keptB.Add more replicas of the handler serviceC.Raise the client timeout to 2 secondsD.Make the independent calls run concurrently so total time is close to the slowest call

← Counting errors with LogQL k6 thresholds →