Skip to main content
Version: 2.6.0

Feature Rollout with Percentile Latency Feedback

Introduction​

This policy rolls out new features based on percentile of latency as the rollout criteria. The percentile of latency is compared with thresholds to determine conditions for advancing, reversing, or resetting the rollout to its initial state. The rollout process consists of a series of steps that progress if the feature is considered healthy.

info

This blueprint is the same as Feature Rollout blueprint, with relevant fields highlighted in the aperturectl generated values file.

Configuration​

Blueprint name: policies/feature-rollout/percentile-latency

Parameters​

Parameterpolicy
DescriptionParameters for the Feature Rollout policy.
TypeObject (policies/feature-rollout/base:schema:rollout_policy)
Default Value
Expand
components: []
drivers:
  percentile_latency_drivers:
    - criteria:
        forward:
          threshold: __REQUIRED_FIELD__
      flux_meter:
        selector:
          - control_point: __REQUIRED_FIELD__
            service: __REQUIRED_FIELD__
        static_buckets:
          buckets:
            - 5
            - 10
            - 25
            - 50
            - 100
            - 250
            - 500
            - 1000
            - 2500
            - 5000
            - 10000
      percentile: 95
evaluation_interval: 10s
load_ramp:
  sampler:
    label_key: ''
    selectors:
      - control_point: __REQUIRED_FIELD__
        service: __REQUIRED_FIELD__
  steps:
    - duration: __REQUIRED_FIELD__
      target_accept_percentage: __REQUIRED_FIELD__
policy_name: __REQUIRED_FIELD__
resources:
  flow_control:
    classifiers: []
rollout: false
Parameterdashboard
DescriptionConfiguration for the Grafana dashboard accompanying this policy.
TypeObject (policies/feature-rollout/base:param:dashboard)
Default Value
Expand
datasource:
  filter_regex: ''
  name: $datasource
extra_filters: {}
refresh_interval: 5s
time_from: now-15m
time_to: now
title: Aperture Feature Rollout

Dynamic Configuration​

note

The following configuration parameters can be dynamically configured at runtime, without reloading the policy.

Parameters​

Parameterpass_through_label_values
DescriptionSpecify certain label values to be always accepted by the _Sampler_ regardless of accept percentage. This configuration can be updated at the runtime without shutting down the policy.
TypeArray of string
Default Value
Expand
- __REQUIRED_FIELD__