2.0.2 - 2022-02-25 - FIX: runner was not requiring unicorn integration correctly leading to a crash 2.0.1 - 2022-02-24 - FIX: ensure threads do not leak when calling #start repeatedly on instrumentation classes, this is an urgent patch for Puma integration 2.0.0 - 2022-02-18 - FEATURE: Add per worker custom labels - FEATURE: support custom histogram buckets - FIX: all metrics are exposing status label, and not only `http_requests_total` - BREAKING: rename all `http_duration` metrics to `http_request_duration` to match prometheus official naming conventions (See https://prometheus.io/docs/practices/naming/#metric-names). 1.0.1 - 2021-12-22 - FEATURE: add labels to preflight requests - FEATURE: SidekiqStats metrics - FIX: mintor refactors to Sidekiq metrics 1.0.0 - 2021-11-23 - BREAKING: rename metrics to match prometheus official naming conventions (See https://prometheus.io/docs/practices/naming/#metric-names) - FEATURE: Sidekiq process metrics - FEATURE: Allow collecting web metrics as histograms - FIX: logger improved for web server - FIX: Remove job labels from DelayedJob queues 0.8.1 - 2021-08-04 - FEATURE: swap from hardcoded STDERR to logger pattern (see README for details) 0.8.0 - 2021-07-05 - FIX: handle ThreadError more gracefully in cases where process shuts down - FEATURE: add job_name and queue_name labels to delayed job metrics - FEATURE: always scope puma metrics on hostname in collector - FEATURE: add customizable labels option to puma collector - FEATURE: support for Resque - DEV: Remove support for EOL ruby 2.5 - FIX: Add source location to MethodProfiler patches - FEATURE: Improve Active Record instrumentation - FEATURE: Support HTTP_X_AMZN_TRACE_ID when supplied 0.7.0 - 2020-12-29 - Dev: Removed support from EOL rubies, only 2.5, 2.6, 2.7 and 3.0 are supported now. - Dev: Better support for Ruby 3.0, explicitly depending on webrick - Dev: Rails 6.1 instrumentation support - FEATURE: clean pattern for overriding middleware labels was introduced (in README) - Fix: Better support for forking 0.6.0 - 2020-11-17 - FEATURE: add support for basic-auth in the prometheus_exporter web server 0.5.3 - 2020-07-29 - FEATURE: added #remove to all metric types so users can remove specific labels if needed 0.5.2 - 2020-07-01 - FEATURE: expanded instrumentation for sidekiq - FEATURE: configurable default labels 0.5.1 - 2020-02-25 - FEATURE: Allow configuring the default client's host and port via environment variables 0.5.0 - 2020-02-14 - Breaking change: listen only to localhost by default to prevent unintended insecure configuration - FIX: Avoid calling `hostname` aggressively, instead cache it on the exporter instance 0.4.17 - 2020-01-13 - FEATURE: add support for `to_h` on all metrics which can be used to query existing key/values 0.4.16 - 2019-11-04 - FEATURE: Support #reset! on all metric types to reset a metric to default 0.4.15 - 2019-11-04 - FEATURE: Improve delayed job collector, add pending counts - FEATURE: New ActiveRecord collector (documented in readme) - FEATURE: Allow passing in histogram and summary options - FEATURE: Allow custom labels for unicorn collector 0.4.14 - 2019-09-10 - FEATURE: allow finding metrics by name RemoteMetric #find_registered_metric - FIX: guard socket closing 0.4.13 - 2019-07-09 - Fix: Memory leak in unicorn and puma collectors 0.4.12 - 2019-05-30 - Fix: unicorn collector reporting incorrect number of unicorn workers 0.4.11 - 2019-05-15 - Fix: Handle stopping nil worker_threads in Client - Dev: add frozen string literals 0.4.10 - 2019-04-29 - Fix: Custom label support for puma collector - Fix: Raindrops socket collector not working correctly 0.4.9 - 2019-04-11 - Fix: Gem was not working correctly in Ruby 2.4 and below due to a syntax error 0.4.8 - 2019-04-10 - Feature: added helpers for instrumenting unicorn using raindrops 0.4.7 - 2019-04-08 - Fix: collector was not escaping " \ and \n correctly. This could lead to a corrupt payload in some cases. 0.4.6 - 2019-04-02 - Feature: Allow resetting a counter - Feature: Add sidekiq metrics: restarted, dead jobs counters - Fix: Client shutting down before sending metrics to collector 0.4.5 - 2019-02-14 - Feature: Allow process collector to ship custom labels for all process metrics - Fix: Always scope process metrics on hostname in collector 0.4.4 - 2019-02-13 - Feature: add support for local metric collection without using HTTP 0.4.3 - 2019-02-11 - Feature: Add alias for Gauge #observe called #set, this makes it a bit easier to migrate from prom - Feature: Add increment and decrement to Counter 0.4.2 - 2018-11-30 - Fix/Feature: setting a Gauge to nil will remove Gauge (setting to non numeric will raise) 0.4.0 - 2018-10-23 - Feature: histogram support - Feature: custom quantile support for summary - Feature: Puma metrics - Fix: delayed job metrics 0.3.4 - 2018-10-02 - Fix: custom collector via CLI was not working correctly 0.3.3 - Feature: Add more metrics to delayed job collector 0.3.2 - Feature: Add posibility to set custom_labels on multi process mode 0.3.1 - Allow runner to accept a --timeout var - Allow runner to accept a blank prefix 0.3.0 - Breaking change: Follow Prometheus metric [naming conventions](https://prometheus.io/docs/practices/naming/#metric-names) 0.1.15 - 2018-02-19 - Feature: Prefer to use oj if it is loadable 0.1.14 - 2018-02-17 - Fix: error when shipping summary metric with no labels - Feature: runner was extracted so it can be reused @304