Skip to content

Feature Request: Add configurable PROMETHEUS_LATENCY_BUCKETS for histogram metrics #21

Description

@jc-67degrees

Feature Request

Add support for configurable latency histogram buckets via a PROMETHEUS_LATENCY_BUCKETS configuration option.

Problem

Currently, tc_prometheus uses Summary metrics for latency tracking with default settings. Users cannot customise histogram bucket boundaries to better match their specific latency profiles and monitoring needs.

For example, if you're serving images with typical response times in the 10-100ms range, the default Prometheus buckets (.005, .01, .025, .05, .1, .25, .5, 1, 2.5, 5, 10 seconds) may not provide optimal granularity for your use case.

Proposed Solution

Add a new configuration option PROMETHEUS_LATENCY_BUCKETS that allows users to specify custom bucket boundaries for latency histograms in their Thumbor configuration:

# Example configuration in thumbor.conf
PROMETHEUS_LATENCY_BUCKETS = [0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1.0, 2.5, 5.0]

Benefits

  • Better visibility into latency distributions specific to each deployment
  • More meaningful percentile calculations (p50, p95, p99)
  • Improved alerting capabilities based on latency SLOs
  • Compatibility with other Prometheus exporters that support configurable buckets

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions