Skip to main content

Prometheus integration: Support for custom defined histogram buckets

Feature RequestIntegrations

When exporting check data through the Prometheus integration I want support for more than just one histogram bucket, so that I have a clearer picture of how my checks perform.

Ideally these would be user defined so that I can adapt them based on the performance of the service I monitor.

Posted by Rikard Nilsson ·
3 comments

Log in to comment and vote

Comments3

  • James Hartig

    •

    May 19, 2025

    I’d like to echo what was already shared. Having a single bucket that’s Infinity is useless since we have no idea what the upper-bound is for the specific metric. Having any other bucket would be more useful but ideally there would be at least 2-3 buckets, we could set some sort of SLO value which would set the bucket, or it would use the new native histogram functionality.

  • Charles Lin

    •

    Feb 18, 2025

    Even better, export histogram that uses the latest native histogram or exponential histogram as it is called in the OpenTelemetry.

  • Anton

    •

    May 29, 2024

    for the context: having just 1 histogram bucket `le="+Inf"` makes very little sense.

    le is a upper inclusive bound. and having it at +Inf results in literally all samples ending up in the bucket, which essentially turns this histogram into a glorified counter.

    even the most basic and opinionated set of buckets would be more useful.