parse-metrics.rb

#!/usr/bin/env ruby

require 'json'

def dig(hash, keys)
  keys.reduce(hash) {|h, k| i = h[k]; i ? i : break }
end

def to_csv(json, fields)
  fields.map{|f| json[f] }.join(',')
end

def get_default_gc_stats(metrics)
  [
    ['jvm.gc.scavenge.count', metrics['PS Scavenge']['count']],
    ['jvm.gc.scavenge.total-time', metrics['PS Scavenge']['total-time-ms']],
    ['jvm.gc.marksweep.count', metrics['PS MarkSweep']['count']],
    ['jvm.gc.marksweep.total-time', metrics['PS MarkSweep']['total-time-ms']]
  ]
end

def get_cms_gc_stats(metrics)
  [
    ['jvm.gc.parnew.count', metrics['ParNew']['count']],
    ['jvm.gc.parnew.total-time', metrics['ParNew']['total-time-ms']],
    ['jvm.gc.marksweep.count', metrics['ConcurrentMarkSweep']['count']],
    ['jvm.gc.marksweep.total-time', metrics['ConcurrentMarkSweep']['total-time-ms']]
  ]
end

def get_g1_gc_stats(metrics)
  [
    ['jvm.gc.new-gen.count', metrics['G1 Young Generation']['count']],
    ['jvm.gc.new-gen.total-time', metrics['G1 Young Generation']['total-time-ms']],
    ['jvm.gc.old-gen.count', metrics['G1 Old Generation']['count']],
    ['jvm.gc.old-gen.total-time', metrics['G1 Old Generation']['total-time-ms']]
  ]
end

JRUBY_METRICS = %w(borrow-timeout-count borrow-retry-count average-borrow-time average-wait-time)
JRUBY_HIERARCHY = %w(jruby-metrics status experimental metrics)
HTTP_HIERARCHY = %w(master status experimental http-metrics)
JVM_HIERARCHY = %w(status-service status experimental jvm-metrics)

HEADER = %w(
  jruby-metrics.borrow-timeout-count
  jruby-metrics.borrow-retry-count
  jruby-metrics.average-borrow-time
  jruby-metrics.average-wait-time
  jvm.heap-memory.committed
  jvm.heap-memory.init
  jvm.heap-memory.used
  jvm.non-heap-memory.committed
  jvm.non-heap-memory.init
  jvm.non-heap-memory.used
  jvm.file-descriptors
  jvm.cpu-usage
  jvm.up-time
  jvm.thread-count
  jvm.gc.scavenge.count
  jvm.gc.scavenge.total-time
  jvm.gc.parnew.count
  jvm.gc.parnew.total-time
  jvm.gc.marksweep.count
  jvm.gc.marksweep.total-time
  jvm.gc.new-gen.count
  jvm.gc.new-gen.total-time
  jvm.gc.old-gen.count
  jvm.gc.old-gen.total-time)


def format_puppet_info(metrics)
  dig(metrics, PUPPET_HIERARCHY)
end

def format_jruby_metrics(metrics)
  dig(metrics, JRUBY_HIERARCHY).
    select {|k, _| JRUBY_METRICS.include? k }.
    map {|k, v| ['jruby-metrics.' + k.to_s, v] }.
    to_h
end

def format_jvm_metrics(raw_metrics)
  metrics = dig(raw_metrics, JVM_HIERARCHY)
  memory = %w(heap-memory non-heap-memory).flat_map {|mem_type|
             %w(committed init used).map {|mem_metric|
               ["jvm.#{mem_type}.#{mem_metric}", metrics[mem_type][mem_metric]]
             }
           }
  system = [
    ['jvm.file-descriptors', metrics['file-descriptors']['used']],
    ['jvm.cpu-usage', metrics['cpu-usage']],
    ['jvm.up-time', metrics['up-time-ms']]
  ]

  if metrics['threading']
    system << ['jvm.thread-count', metrics['threading']['thread-count']]
  end

  if metrics['gc-stats']['PS Scavenge']
    gc = get_default_gc_stats(metrics['gc-stats'])
  elsif metrics['gc-stats']['ParNew']
    gc = get_cms_gc_stats(metrics['gc-stats'])
  elsif metrics['gc-stats']['G1 Young Generation']
    gc = get_g1_gc_stats(metrics['gc-stats'])
  end

  (gc + system + memory).to_h
end

out = File.open(ARGV[0]).each_line.map do |line|
  raw_metrics = JSON.parse line
  jruby_metrics = format_jruby_metrics(raw_metrics)
  jvm_metrics = format_jvm_metrics(raw_metrics)

  metrics = jruby_metrics.merge(jvm_metrics)
  HEADER.map {|field| metrics[field] }.join(',')
end

STDOUT.puts HEADER.join(',')
STDOUT.puts out