prometheus-client-mmap 0.25.0 → 1.1.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 15dde2bfa3d25ee344e811944e599f0a64ed199caf949310cbfc99a7af4ccfd3
-  data.tar.gz: 6bd6f3a9d0d7c055b1138acfe06f6d206edeb73481eff5be82079ed36c73c64e
+  metadata.gz: 971eb514c45dcdbf4b312b1b0ea74b3060e8879085e9947da5af0b658e711f83
+  data.tar.gz: 69f018cddf0684a280d70ed06cf8c2a04c6f5c0fd3fad749b9624185eb877584
 SHA512:
-  metadata.gz: 23e114d89e09c771308e877ee0218aa94daf01df55ff0c29875a41281b762ecc09e831b0e5bafcaf7aa5b9b6a67d8e95328bbeaf2ddad867b6901a301650a068
-  data.tar.gz: 0e1c4b6bce251e33c8551eba06b5b1167cdd3e56eaaac9caee7f9256546c83430a66e8d7481bad550e52bb6690810e31f688379a65792e1d81d251b36a8958dd
+  metadata.gz: ae066f96a7585fdd96ef7604d5749c56996176826eeb834765c1375f734c60838e94fa867b8a333bbc48a0ae50917f6096efc7e52ed2054564bf620f1cd71384
+  data.tar.gz: 7738eeef01025906e0897b08ee41e5f26a65f781b71374fbd9aed2cfba6eea5b62735c789ed0cb879ca10a1936d72a967709f2ca9f63c7fe0a996df3c7291cf7

data/.tool-versions

@@ -1 +1 @@
-rust 1.65.0
+rust 1.73.0

data/README.md

@@ -10,7 +10,24 @@ through a HTTP interface. Intended to be used together with a
 
 [![Gem Version][4]](http://badge.fury.io/rb/prometheus-client-mmap)
 [![Build Status][3]](https://gitlab.com/gitlab-org/prometheus-client-mmap/commits/master)
-[![Dependency Status][5]](https://gemnasium.com/prometheus/prometheus-client-mmap)
+
+## Installation
+
+prometheus-client-mmap ships with [precompiled native gems](https://rubygems.org/gems/prometheus-client-mmap).
+Install the gem via:
+
+```shell
+gem install prometheus-client-mmap
+```
+
+### Requirements for building from source
+
+Building from source requires a number of dependencies:
+
+* Rust v1.65+
+* clang 5.0 or higher for [bindgen](https://rust-lang.github.io/rust-bindgen/requirements.html)
+* `make`
+* A C compiler (for legacy extension, which will be removed soon)
 
 ## Usage
 
@@ -34,19 +51,11 @@ http_requests = prometheus.counter(:http_requests, 'A counter of HTTP requests m
 http_requests.increment
 ```
 
-## Rust extension (experimental)
+## Rust extension
 
-In an effort to improve maintainability, there is now an optional Rust
+This gem now uses a rewritten Rust extension instead of C.
 implementation that reads the metric files and outputs the multiprocess
-metrics to text. If `rustc` is available, then the Rust extension will
-be built automatically. The `use_rust` keyword argument can be used:
-
-```ruby
-puts Prometheus::Client::Formats::Text.marshal_multiprocess(use_rust: true)
-```
-
-Note that this parameter will likely be deprecated and removed once the Rust
-extension becomes the default mode.
+metrics to text. This implementation is significantly faster than the C extension.
 
 ### Rack middleware
 
@@ -200,23 +209,12 @@ Set `prometheus_multiproc_dir` environment variable to the path where you want m
 prometheus_multiproc_dir=/tmp
 ```
 
-### Multiprocess metrics via Rust extension
-
-If the environment variable `prometheus_rust_multiprocess_metrics=true` is set or if the `rust_multiprocess_metrics`
-configuration setting is `true` and the `fast_mmaped_file_rs` extension is available, it will be used to generate
-multiprocess metrics. This should be significantly faster than the C extension.
-
-### Read and write metrics via Rust extension
-
-If the environment variable `prometheus_rust_mmaped_file=true` is set then if the `fast_mmaped_file_rs`
-extension is available it will be used to read and write metrics from the mmapped file.
-
 ## Pitfalls
 
 ### PID cardinality
 
-In multiprocess setup e.g. running under Unicorn, having worker process restart often can
-lead to performance problems when proccesing metric files. By default each process using
+In multiprocess setup e.g. running under Unicorn or Puma, having worker process restart often
+can lead to performance problems when proccesing metric files. By default each process using
 Prometheus metrics will create a set of files based on that process PID. With high worker
 churn this will lead to creation of thousands of files and in turn will cause very noticable
 slowdown when displaying metrics
@@ -224,17 +222,23 @@ slowdown when displaying metrics
 To reduce this problem, a surrogate process id can be used. Set of all such IDs needs
 have low cardinality, and each process id must be unique among all running process.
 
-For Unicorn a worker id/number can be used to greatly speedup the metrics rendering.
+For Unicorn and Puma a worker id/number can be used to greatly speedup the metrics rendering.
 
-To use it add this line to your `configure` block:
+If you are using Unicorn, add this line to your `configure` block:
 
 ```ruby
   config.pid_provider = Prometheus::Client::Support::Unicorn.method(:worker_pid_provider)
 ```
 
+If you are using Puma, add this line to your `configure` block:
+
+```ruby
+  config.pid_provider = Prometheus::Client::Support::Puma.method(:worker_pid_provider)
+```
+
 ## Tools
 
-###`bin/parse`
+### `bin/parse`
 
 This command can be used to parse metric files located on the filesystem just like a metric exporter would.
 It outputs either `json` formatted raw data or digested data in prometheus `text` format.