pglombardo

NameError in Refinery::Admin::CoreController#index
uninitialized constant Refinery::I18n::Engine::Globalize

Rails.root: /home/pglombardo/Projects/ruby-test-stacks/refinethis2

Application Trace | Framework Trace | Full Trace
refinerycms-i18n (3.0.1) lib/refinery/i18n/engine.rb:63:in `globalize!'
activesupport (4.1.9) lib/active_support/callbacks.rb:424:in `block in make_lambda'
activesupport (4.1.9) lib/active_support/callbacks.rb:160:in `call'
activesupport (4.1.9) lib/active_support/callbacks.rb:160:in `block in halting'

pglombardo

workers Integer(ENV['WEB_CONCURRENCY'] || 2)
threads_count = Integer(ENV['MAX_THREADS'] || 5)
threads threads_count, threads_count

preload_app!

rackup      DefaultRackup
port        ENV['PORT']     || 3000
environment ENV['RACK_ENV'] || 'development'

pglombardo

#!/usr/bin/env ruby

Bundler.require(:default)

Oboe::Config[:tracing_mode] = :always

while true do
  Oboe.sample?(:layer => 'rack')
  puts "PID - SampleRate/Source: #{Process.pid} - #{Oboe.sample_rate}/#{Oboe.sample_source}"
  sleep 1

pglombardo

GC.disable

Wall Clock Time versus CPU Time

An important difference to note is the how time is reported by various measurement methods. Wall clock time is the actual time passed in terms of human perception whereas CPU time is the time spent processing the work. CPU time doesn't include any delays waiting on resources to free up such as thread interrupts or garbage collection.

The Work to Measure

To keep things simple, we'll create a Ruby Proc and just repeatedly call that Proc for each of the measurement methods below.

pglombardo

      def report_backtrace
        kvs = { :Backtrace => Oboe::API.backtrace }

        if Oboe.tracing?
          Oboe::API.log(nil, 'info', kvs)
        else
          Oboe::API.start_trace('manual', { "Force" => true }) do
            Oboe::API.log(nil, 'info', kvs)
          end
        end

pglombardo

EventMachine::HttpConnection.class_eval do
  def setup_request_with_oboe(*args, &block)
    c = nil

    ::Oboe::API.log_entry('em-http-request', { uri: @uri, start: 'start' })

    c = setup_request_without_oboe(*args, &block)
    c.req.headers["X-Trace"] = Oboe::Context.toString()
    ::Oboe::API.log(nil, 'info', { uri: @uri })

pglombardo

HOW-TO: ASYNCHRONOUS AND PARALLEL WORKFLOWS

Most web application requests follow a simple path through the application, regardless of whether it is blocking or evented, where the critical path of the request is the only work being done. However, some web applications make use of threads, eventlets, or queues to process work in parallel or out of the critical path.

In order to analyze traces with asynchronous and parallel workflows correctly, small annotations must be added to the trace’s events. They boil down to adding the key Async = True on “asynchronous” layers. Here’s how to handle two common scenarios:

pglombardo

@session = Moped::Session.new([ "127.0.0.1:27017" ])
@session.use :moped_test
@users = @session[:users]

Oboe::API.start_trace('just_mongo_find', nil, { :Force => true }) do
  @users.find
end

Oboe::API.start_trace('mongo_find_and_count', nil, { :Force => true }) do
  @users.find.count

pglombardo

IRC2HipChat is a short Ruby script that uses Cinch to monitor IRC and re-post messages to a dedicated HipChat room using the HipChat API gem.

  on :channel do |m|
    msg = "<em>#{m.user.nick}</em>: #{m.message}"
    $hipchat.send('irc2hipchat', msg, { :notify => true, :message_format => 'html' })
  end

pglombardo

require 'dashing'
require 'oboe'

Oboe::Config[:tracing_mode] = 'always'
Oboe::Config[:sample_rate] = 1e6
Oboe::Config[:verbose] = true

configure do
  set :auth_token, 'YOUR_AUTH_TOKEN'