Skip to content
Snippets Groups Projects
Select Git revision
  • move-gl-dropdown
  • improve-table-pagination-spec
  • move-markdown-preview
  • winh-fix-merge-request-spec
  • master default
  • index-namespaces-lower-name
  • winh-single-karma-test
  • 10-3-stable
  • 36782-replace-team-user-role-with-add_role-user-in-specs
  • winh-modal-internal-state
  • tz-ide-file-icons
  • 38869-milestone-select
  • update-autodevops-template
  • jivl-activate-repo-cookie-preferences
  • qa-add-deploy-key
  • docs-move-article-ldap
  • 40780-choose-file
  • 22643-manual-job-page
  • refactor-cluster-show-page-conservative
  • dm-sidekiq-versioning
  • v10.4.0.pre
  • v10.3.0
  • v10.3.0-rc5
  • v10.3.0-rc4
  • v10.3.0-rc3
  • v10.3.0-rc2
  • v10.2.5
  • v10.3.0-rc1
  • v10.0.7
  • v10.1.5
  • v10.2.4
  • v10.2.3
  • v10.2.2
  • v10.2.1
  • v10.3.0.pre
  • v10.2.0
  • v10.2.0-rc4
  • v10.2.0-rc3
  • v10.1.4
  • v10.2.0-rc2
40 results

system.rb

Forked from GitLab.org / GitLab FOSS
20811 commits behind the upstream repository.
  • Yorick Peterse's avatar
    d7b4f36a
    Use clock_gettime for all performance timestamps · d7b4f36a
    Yorick Peterse authored
    Process.clock_gettime allows getting the real time in nanoseconds as
    well as allowing one to get a monotonic timestamp. This offers greater
    accuracy without the overhead of having to allocate a Time instance. In
    general using Time.now/Time.new is about 2x slower than using
    Process.clock_gettime(). For example:
    
        require 'benchmark/ips'
    
        Benchmark.ips do |bench|
          bench.report 'Time.now' do
            Time.now.to_f
          end
    
          bench.report 'clock_gettime' do
            Process.clock_gettime(Process::CLOCK_MONOTONIC, :millisecond)
          end
    
          bench.compare!
        end
    
    Running this benchmark gives:
    
        Calculating -------------------------------------
                    Time.now   108.052k i/100ms
               clock_gettime   125.984k i/100ms
        -------------------------------------------------
                    Time.now      2.343M (± 7.1%) i/s -     11.670M
               clock_gettime      4.979M (± 0.8%) i/s -     24.945M
    
        Comparison:
               clock_gettime:  4979393.8 i/s
                    Time.now:  2342986.8 i/s - 2.13x slower
    
    Another benefit of using Process.clock_gettime() is that we can simplify
    the code a bit since it can give timestamps in nanoseconds out of the
    box.
    Use clock_gettime for all performance timestamps
    Yorick Peterse authored
    Process.clock_gettime allows getting the real time in nanoseconds as
    well as allowing one to get a monotonic timestamp. This offers greater
    accuracy without the overhead of having to allocate a Time instance. In
    general using Time.now/Time.new is about 2x slower than using
    Process.clock_gettime(). For example:
    
        require 'benchmark/ips'
    
        Benchmark.ips do |bench|
          bench.report 'Time.now' do
            Time.now.to_f
          end
    
          bench.report 'clock_gettime' do
            Process.clock_gettime(Process::CLOCK_MONOTONIC, :millisecond)
          end
    
          bench.compare!
        end
    
    Running this benchmark gives:
    
        Calculating -------------------------------------
                    Time.now   108.052k i/100ms
               clock_gettime   125.984k i/100ms
        -------------------------------------------------
                    Time.now      2.343M (± 7.1%) i/s -     11.670M
               clock_gettime      4.979M (± 0.8%) i/s -     24.945M
    
        Comparison:
               clock_gettime:  4979393.8 i/s
                    Time.now:  2342986.8 i/s - 2.13x slower
    
    Another benefit of using Process.clock_gettime() is that we can simplify
    the code a bit since it can give timestamps in nanoseconds out of the
    box.
system.rb 1.71 KiB
module Gitlab
  module Metrics
    # Module for gathering system/process statistics such as the memory usage.
    #
    # This module relies on the /proc filesystem being available. If /proc is
    # not available the methods of this module will be stubbed.
    module System
      if File.exist?('/proc')
        # Returns the current process' memory usage in bytes.
        def self.memory_usage
          mem   = 0
          match = File.read('/proc/self/status').match(/VmRSS:\s+(\d+)/)

          if match and match[1]
            mem = match[1].to_f * 1024
          end

          mem
        end

        def self.file_descriptor_count
          Dir.glob('/proc/self/fd/*').length
        end
      else
        def self.memory_usage
          0.0
        end

        def self.file_descriptor_count
          0
        end
      end

      # THREAD_CPUTIME is not supported on OS X
      if Process.const_defined?(:CLOCK_THREAD_CPUTIME_ID)
        def self.cpu_time
          Process.
            clock_gettime(Process::CLOCK_THREAD_CPUTIME_ID, :millisecond).to_f
        end
      else
        def self.cpu_time
          Process.
            clock_gettime(Process::CLOCK_PROCESS_CPUTIME_ID, :millisecond).to_f
        end
      end

      # Returns the current real time in a given precision.
      #
      # Returns the time as a Float.
      def self.real_time(precision = :millisecond)
        Process.clock_gettime(Process::CLOCK_REALTIME, precision).to_f
      end

      # Returns the current monotonic clock time in a given precision.
      #
      # Returns the time as a Float.
      def self.monotonic_time(precision = :millisecond)
        Process.clock_gettime(Process::CLOCK_MONOTONIC, precision).to_f
      end
    end
  end
end