You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have started experimenting with native histogram, although we have same memory but i have noticed the percentile values are not accurate anymore. is there a way to make more accurate by increasing number of buckets or something ?
The text was updated successfully, but these errors were encountered:
Could you bring more details around your accuracy issues with Native Histogram mode, please?
For instance, when you refer to the percentile values, do you mean the ones displayed in the dashboard, directly from Prometheus? And, when you say "are not accurate anymore", what are you comparing them against to? Any examples?
Moreover, considering you have enabled Native Histogram at Prometheus level as well, is that something you experience also with other metrics, not reported by k6? Or do you only experience such loss of accuracy on metrics sent by k6?
cc/ @codebien as I think you implemented that support, so you may know what could be going wrong here.
Thanks! 🙇🏻
codebien
transferred this issue from grafana/xk6-output-prometheus-remote
Feb 21, 2025
I have started experimenting with native histogram, although we have same memory but i have noticed the percentile values are not accurate anymore. is there a way to make more accurate by increasing number of buckets or something ?
The text was updated successfully, but these errors were encountered: