sensu-plugins-mesos 0.1.1 → 1.0.0

checksums.yaml

@@ -1,15 +1,7 @@
 ---
-!binary "U0hBMQ==":
-  metadata.gz: !binary |-
-    YWRlMTBkZmJmZjJiNDY0YmU5MWQxY2JhMDZiZWY5M2Q1NjhkMTE3ZQ==
-  data.tar.gz: !binary |-
-    YzMwZWNjZmU0OTRjYTU5NmZjZTYzYzVlNjViMzRlN2QyOGYzNjEwMg==
+SHA1:
+  metadata.gz: 21411bface3e1d88fc6c4041e0b5dac80044a364
+  data.tar.gz: 806ab66415a37de8c8e724078d74ca66696607ca
 SHA512:
-  metadata.gz: !binary |-
-    MWYwYTY2NzE0MjE0ZTJkNGU5ODRlZjI5MmM4OTFjYjg1YTVlOTYwODk2MTlh
-    MzY1NjM5MGVkMGU1YzVkNjcyOTg2ZDc4MGZkOWU5MTEyMTI2OThjMmVkZWJm
-    MTQ0ZDU3N2Y4ZGJjNTJiMzE0N2I5MmE0ZTJhNDJmYWRjNDdhNTM=
-  data.tar.gz: !binary |-
-    MTc3N2U5OWFiM2E1NTI3Njk4MDU1NzQyMDIyNWRlY2E5NzJjZTJmMzhlNTUy
-    YTVkMTI1ZmFiZTliOGExM2FmNTg0NzA5MGQxNmQ0YWQ0MDVjMGVlNjM3NGZm
-    ZDliYzNjMTdhYzQwY2U5OTdjZTk4MjJiODg1ODk1ZDlkYTNkYmE=
+  metadata.gz: c5a02b94924115f6fed7d4fc57d4cd960624c0caf674b834375cf1733ba1275105c6cb9af26c6893eac04e95a5ca8cde46516fa0ebc390fd17e7118132653fd9
+  data.tar.gz: fab71f56c0672988d61d098e40e61ae47ebcdeca229e61588f02877630b57d72643cefe493e5298f7e46d22d7cc171d0aa4de54a6ab28257dfa9dbac305096eb

data/CHANGELOG.md

@@ -4,6 +4,26 @@ This project adheres to [Semantic Versioning](http://semver.org/).
 This CHANGELOG follows the format listed at [Keep A Changelog](http://keepachangelog.com/)
 
 ## [Unreleased]
+## [1.0.0] - 2017-05-05
+### Breaking Change
+- check-mesos.rb: removed the `mode` parameter (@luisdavim)
+
+### Added
+- check-metronome.rb: Check if Metronome is running (@luisdavim)
+- check-mesos-cpu-balance.rb: Check for imballanced use of CPU accross mesos agents (@luisdavim)
+- check-mesos-gpu-balance.rb: Check for imballanced use of GPU accross mesos agents (@luisdavim)
+- check-mesos-disk-balance.rb: Check for imballanced use of disk accross mesos agents (@luisdavim)
+- check-mesos-mem-balance.rb: Check for imballanced use of memory accross mesos agents (@luisdavim)
+
+### Changed
+- check-marathon-task.rb: Use the health check results to verify that a task is running. (@andrelaszlo)
+- check-marathon-task.rb: Rename incorrect "state" parameter to "status". (@andrelaszlo)
+- Add https support and authentication to marathon plugins: (thanks to Erasys GmbH)
+    - Add "protocol" option to check-marathon and metrics-marathon
+    - Add "protocol", "username" and "password" options to check-marathon-task
+- All checks now have a configurable API endpoint using --uri or -u (@luisdavim)
+- Support the latest Mesos API (@luisdavim)
+- Dropped support for Ruby 1.9.3 (@luisdavim)
 
 ## [0.1.1] - 2016-03-04
 ### Added
@@ -41,7 +61,8 @@ This CHANGELOG follows the format listed at [Keep A Changelog](http://keepachang
 ### Added
 - initial release
 
-[Unreleased]: https://github.com/sensu-plugins/sensu-plugins-mesos/compare/0.1.1...HEAD
+[Unreleased]: https://github.com/sensu-plugins/sensu-plugins-mesos/compare/1.0.0...HEAD
+[1.0.0]: https://github.com/sensu-plugins/sensu-plugins-mesos/compare/0.1.1...1.0.0
 [0.1.1]: https://github.com/sensu-plugins/sensu-plugins-mesos/compare/0.1.0...0.1.1
 [0.1.0]: https://github.com/sensu-plugins/sensu-plugins-mesos/compare/0.0.4...0.1.0
 [0.0.4]: https://github.com/sensu-plugins/sensu-plugins-mesos/compare/0.0.2...0.0.4

data/README.md

@@ -10,8 +10,13 @@
 
 ## Files
  * bin/check-chronos.rb
+ * bin/check-metronome.rb
  * bin/check-marathon.rb
  * bin/check-mesos.rb
+ * bin/check-mesos-cpu-balance.rb
+ * bin/check-mesos-disk-balance.rb
+ * bin/check-mesos-gpu-balance.rb
+ * bin/check-mesos-mem-balance.rb
  * bin/metrics-marathon.rb
  * bin/metrics-mesos.rb
 

data/bin/check-chronos.rb

@@ -42,6 +42,12 @@ class ChronosNodeStatus < Sensu::Plugin::Check::CLI
          long: '--port PORT',
          default: '80'
 
+  option :uri,
+         description: 'Endpoint URI',
+         short: '-u URI',
+         long: '--uri URI',
+         default: '/scheduler/jobs'
+
   option :timeout,
          description: 'timeout in seconds',
          short: '-t TIMEOUT',
@@ -51,10 +57,11 @@ class ChronosNodeStatus < Sensu::Plugin::Check::CLI
 
   def run
     servers = config[:server]
+    uri = config[:uri]
     failures = []
     servers.split(',').each do |server|
       begin
-        r = RestClient::Resource.new("http://#{server}:#{config[:port]}/scheduler/jobs", timeout: config[:timeout]).get
+        r = RestClient::Resource.new("http://#{server}:#{config[:port]}#{uri}", timeout: config[:timeout]).get
         if r.code != 200
           failures << "Chronos on #{server} is not responding"
         end

data/bin/check-marathon-task.rb

@@ -33,48 +33,144 @@ require 'sensu-plugin/check/cli'
 require 'net/http'
 require 'json'
 
+# This plugin checks that the given Mesos/Marathon task is running properly.
+#
+# This means that all of the following is true:
+# 1. There are N tasks for the app, as defined by the --instances parameter
+# 2. Each task's state is running
+# 3. No task is unhealthy, as defined in Marathon
+#
+# A task is seen as **unhealthy** by Marathon if any of the health checks for
+# the task is not **alive**.  Alive means that a check has a last success that
+# is more recent than last failure. It's not alive if the last failure is more
+# recent than the last success, or if the last success doesn't exist at all.
 class MarathonTaskCheck < Sensu::Plugin::Check::CLI
   check_name 'CheckMarathonTask'
 
-  option :server, short: '-s SERVER', long: '--server SERVER', required: true
-  option :port, short: '-p PORT', long: '--port PORT', default: 8080
-  option :task, short: '-t TASK', long: '--task TASK', required: true
-  option :instances, short: '-i INSTANCES', long: '--instances INSTANCES', required: true, proc: proc(&:to_i)
+  option :server,
+         short: '-s SERVER',
+         long: '--server SERVER',
+         required: true
+
+  option :port,
+         short: '-p PORT',
+         long: '--port PORT',
+         default: 8080
+
+  option :uri,
+         description: 'Endpoint URI',
+         short: '-u URI',
+         long: '--uri URI',
+         default: '/v2/tasks?status=running'
+
+  option :task,
+         short: '-t TASK',
+         long: '--task TASK',
+         required: true
+
+  option :instances,
+         short: '-i INSTANCES',
+         long: '--instances INSTANCES',
+         required: true,
+         proc: proc(&:to_i)
+
+  option :protocol,
+         short: '-P PROTOCOL',
+         long: '--protocol PROTOCOL',
+         required: false,
+         default: 'http'
+
+  option :username,
+         short: '-u USERNAME',
+         long: '--username USERNAME',
+         required: false
+
+  option :password,
+         long: '--password PASSWORD',
+         required: false
 
   def run
-    if config[:instances] == 0
+    if config[:instances].zero?
       unknown 'number of instances should be an integer'
     end
 
+    if !config[:username].nil? && config[:password].nil? ||
+       config[:username].nil? && !config[:password].nil?
+      unknown 'You must provide both username and password'
+    end
+
     failures = []
+    uri = config[:uri]
     config[:server].split(',').each do |s|
       begin
-        url = URI.parse("http://#{s}:#{config[:port]}/v2/tasks?state=running")
+        url = URI.parse("#{config[:protocol]}://#{s}:#{config[:port]}#{uri}")
         req = Net::HTTP::Get.new(url)
         req.add_field('Accept', 'application/json')
-        r = Net::HTTP.new(url.host, url.port).start do |h|
+        if !config[:username].nil? && !config[:password].nil?
+          req.basic_auth(config[:username], config[:password])
+        end
+        r = Net::HTTP.start(url.host, url.port,
+                            use_ssl: config[:protocol] == 'https') do |h|
           h.request(req)
         end
 
-        tasks = JSON.parse(r.body)['tasks']
-        tasks.select! do |t|
-          t['appId'] == "/#{config[:task]}"
-        end
+        ok_count, unhealthy = check_tasks r.body
+
+        message = "#{ok_count}/#{config[:instances]} #{config[:task]} tasks running"
 
-        message = "#{tasks.length}/#{config[:instances]} #{config[:task]} tasks running"
+        if unhealthy.any?
+          message << ":\n" << unhealthy.join("\n")
+        end
 
-        if tasks.length < config[:instances]
+        if unhealthy.any? || ok_count < config[:instances]
           critical message
         end
 
         ok message
       rescue Errno::ECONNREFUSED, SocketError
         failures << "Marathon on #{s} could not be reached"
-      rescue
-        failures << "error caught trying to reach Marathon on #{s}"
+      rescue => err
+        failures << "error caught trying to reach Marathon on #{s}: #{err}"
       end
     end
 
     unknown "marathon task state could not be retrieved:\n" << failures.join("\n")
   end
+
+  # Parses JSON data as returned from Marathon's tasks API
+  # @param data [String] Server response
+  # @return [Numeric, [String]] Number of running tasks and a list of error
+  #                             messages from unhealthy tasks
+  def check_tasks(data)
+    begin
+      tasks = JSON.parse(data)['tasks']
+    rescue JSON::ParserError
+      raise "Could not parse JSON response: #{data}"
+    end
+
+    if tasks.nil?
+      raise "No tasks in server response: #{data}"
+    end
+
+    tasks.select! do |t|
+      t['appId'] == "/#{config[:task]}"
+    end
+
+    unhealthy = []
+
+    # Collect last error message for all health checks that are not alive
+    tasks.each do |task|
+      checks = task['healthCheckResults'] || []
+      checks.each do |check|
+        if check['alive']
+          next
+        end
+        message = check['lastFailureCause'] ||
+                  'Health check not alive'
+        unhealthy << message
+      end
+    end
+
+    [tasks.length, unhealthy]
+  end
 end

data/bin/check-marathon.rb

@@ -43,6 +43,19 @@ class MarathonNodeStatus < Sensu::Plugin::Check::CLI
          required: false,
          default: '8080'
 
+  option :protocol,
+         description: 'Marathon protocol [http/https]',
+         short: '-P PROTOCOL',
+         long: '--protocol PROTOCOL',
+         required: false,
+         default: 'http'
+
+  option :uri,
+         description: 'Endpoint URI',
+         short: '-u URI',
+         long: '--uri URI',
+         default: '/ping'
+
   option :timeout,
          description: 'timeout in seconds',
          short: '-t TIMEOUT',
@@ -55,7 +68,7 @@ class MarathonNodeStatus < Sensu::Plugin::Check::CLI
     failures = []
     servers.split(',').each do |server|
       begin
-        r = RestClient::Resource.new("http://#{server}:#{config[:port]}/ping", timeout: config[:timeout]).get
+        r = RestClient::Resource.new("#{config[:protocol]}://#{server}:#{config[:port]}#{config[:uri]}", timeout: config[:timeout]).get
         if r.code != 200
           failures << "Marathon Service on #{server} is not responding"
         end

data/bin/check-mesos-cpu-balance.rb

@@ -0,0 +1,154 @@
+#! /usr/bin/env ruby
+#
+#   check-mesos-cpu-balance
+#
+# DESCRIPTION:
+#   This plugin checks that there is less CPU imbalance than specified on a certain mesos cluster
+#
+# OUTPUT:
+#   plain text
+#
+# PLATFORMS:
+#   Linux
+#
+# DEPENDENCIES:
+#   gem: sensu-plugin
+#   gem: rest-client
+#   gem: json
+#
+# USAGE:
+#   #YELLOW
+#
+# NOTES:
+#
+# LICENSE:
+#   Released under the same terms as Sensu (the MIT license); see LICENSE
+#   for details.
+#
+
+require 'sensu-plugin/check/cli'
+require 'rest-client'
+require 'json'
+
+class MesosCpuBalanceCheck < Sensu::Plugin::Check::CLI
+  check_name 'MesosCpuBalanceCheck'
+  @metrics_name = 'slaves'.freeze
+  CHECK_TYPE = 'cpus'.freeze
+
+  class << self
+    attr_reader :metrics_name
+  end
+
+  option :server,
+         description: 'Mesos server',
+         short: '-s SERVER',
+         long: '--server SERVER',
+         default: 'localhost'
+
+  option :port,
+         description: 'port (default 5050)',
+         short: '-p PORT',
+         long: '--port PORT',
+         default: 5050,
+         required: false
+
+  option :uri,
+         description: 'Endpoint URI',
+         short: '-u URI',
+         long: '--uri URI',
+         default: '/master/slaves'
+
+  option :timeout,
+         description: 'timeout in seconds',
+         short: '-t TIMEOUT',
+         long: '--timeout TIMEOUT',
+         proc: proc(&:to_i),
+         default: 5
+
+  option :crit,
+         description: 'Critical value to check against',
+         short: '-c VALUE',
+         long: '--critical VALUE',
+         proc: proc(&:to_i),
+         default: 0,
+         required: false
+
+  option :warn,
+         description: 'Warning value to check against',
+         short: '-w VALUE',
+         long: '--warning VALUE',
+         proc: proc(&:to_i),
+         default: 0,
+         required: false
+
+  def run
+    if config[:crit] < 0 || config[:warn] < 0
+      unknown "Thresholds cannot be negative, crit: #{config[:crit]}, warn: #{config[:warn]}"
+    end
+
+    server = config[:server]
+    port = config[:port]
+    uri = config[:uri]
+    timeout = config[:timeout]
+    crit = config[:crit]
+    warn = config[:warn]
+
+    begin
+      server = get_leader_url server, port
+      r = RestClient::Resource.new("#{server}#{uri}", timeout).get
+      compare = get_check_diff(get_slaves(r))
+      if compare['diff'] >= crit
+        critical "There is a CPU diff of #{compare['diff']} bigger than #{crit} " + compare['msg']
+      end
+      if compare['diff'] >= warn
+        warning "There is a CPU diff of #{compare['diff']} bigger than #{warn} " + compare['msg']
+      end
+    rescue Errno::ECONNREFUSED, RestClient::ResourceNotFound, SocketError
+      unknown  "Mesos #{server} is not responding"
+    rescue RestClient::RequestTimeout
+      unknown  "Mesos #{server} connection timed out"
+    end
+    ok
+  end
+
+  # Redirects server call to discover the Leader
+  # @param server [String] Server address
+  # @param port [Number] api port
+  # @return [Url] Url representing the Leader
+
+  def get_leader_url(server, port)
+    RestClient::Resource.new("http://#{server}:#{port}/redirect").get.request.url
+  end
+
+  # Parses JSON data as returned from Mesos's metrics API
+  # @param data [String] Server response
+  # @return [Integer] Number of failed tasks in Mesos
+  def get_slaves(data)
+    begin
+      slaves = JSON.parse(data)[MesosCpuBalanceCheck.metrics_name]
+    rescue JSON::ParserError
+      raise "Could not parse JSON response: #{data}"
+    end
+
+    if slaves.nil?
+      raise "No metrics for [#{MesosCpuBalanceCheck.metrics_name}] in server response: #{data}"
+    end
+
+    slaves
+  end
+
+  def get_check_diff(slavelist)
+    begin
+      usages = {}
+      check_diff = {}
+      slavelist.each do |slaveinfo|
+        usages.store(slaveinfo['hostname'], slaveinfo['used_resources'][CHECK_TYPE] * 100 / slaveinfo['resources'][CHECK_TYPE])
+      end
+      sorted = usages.sort_by { |_hostname, total| total }
+      max = usages.length - 1
+      check_diff['diff'] = sorted[max][1] - sorted[0][1]
+      check_diff['msg'] = "Hostname #{sorted[0][0]} uses #{sorted[0][1]}% and Hostname #{sorted[max][0]} uses #{sorted[max][1]}%"
+    end
+    check_diff
+  end
+end

data/bin/check-mesos-disk-balance.rb

@@ -0,0 +1,154 @@
+#! /usr/bin/env ruby
+#
+#   check-mesos-disk-balance
+#
+# DESCRIPTION:
+#   This plugin checks that there is less disk usage imbalance than specified on a certain mesos cluster
+#
+# OUTPUT:
+#   plain text
+#
+# PLATFORMS:
+#   Linux
+#
+# DEPENDENCIES:
+#   gem: sensu-plugin
+#   gem: rest-client
+#   gem: json
+#
+# USAGE:
+#   #YELLOW
+#
+# NOTES:
+#
+# LICENSE:
+#   Released under the same terms as Sensu (the MIT license); see LICENSE
+#   for details.
+#
+
+require 'sensu-plugin/check/cli'
+require 'rest-client'
+require 'json'
+
+class MesosDiskBalanceCheck < Sensu::Plugin::Check::CLI
+  check_name 'MesosDiskBalanceCheck'
+  @metrics_name = 'slaves'.freeze
+  CHECK_TYPE = 'disk'.freeze
+
+  class << self
+    attr_reader :metrics_name
+  end
+
+  option :server,
+         description: 'Mesos server',
+         short: '-s SERVER',
+         long: '--server SERVER',
+         default: 'localhost'
+
+  option :port,
+         description: 'port (default 5050)',
+         short: '-p PORT',
+         long: '--port PORT',
+         default: 5050,
+         required: false
+
+  option :uri,
+         description: 'Endpoint URI',
+         short: '-u URI',
+         long: '--uri URI',
+         default: '/master/slaves'
+
+  option :timeout,
+         description: 'timeout in seconds',
+         short: '-t TIMEOUT',
+         long: '--timeout TIMEOUT',
+         proc: proc(&:to_i),
+         default: 5
+
+  option :crit,
+         description: 'Critical value to check against',
+         short: '-c VALUE',
+         long: '--critical VALUE',
+         proc: proc(&:to_i),
+         default: 0,
+         required: false
+
+  option :warn,
+         description: 'Warning value to check against',
+         short: '-w VALUE',
+         long: '--warning VALUE',
+         proc: proc(&:to_i),
+         default: 0,
+         required: false
+
+  def run
+    if config[:crit] < 0 || config[:warn] < 0
+      unknown "Thresholds cannot be negative, crit: #{config[:crit]}, warn: #{config[:warn]}"
+    end
+
+    server = config[:server]
+    port = config[:port]
+    uri = config[:uri]
+    timeout = config[:timeout]
+    crit = config[:crit]
+    warn = config[:warn]
+
+    begin
+      server = get_leader_url server, port
+      r = RestClient::Resource.new("#{server}#{uri}", timeout).get
+      compare = get_check_diff(get_slaves(r))
+      if compare['diff'] >= crit
+        critical "There is a disk usage diff of #{compare['diff']} bigger than #{crit} " + compare['msg']
+      end
+      if compare['diff'] >= warn
+        warning "There is a disk usage diff of #{compare['diff']} bigger than #{warn} " + compare['msg']
+      end
+    rescue Errno::ECONNREFUSED, RestClient::ResourceNotFound, SocketError
+      unknown  "Mesos #{server} is not responding"
+    rescue RestClient::RequestTimeout
+      unknown  "Mesos #{server} connection timed out"
+    end
+    ok
+  end
+
+  # Redirects server call to discover the Leader
+  # @param server [String] Server address
+  # @param port [Number] api port
+  # @return [Url] Url representing the Leader
+
+  def get_leader_url(server, port)
+    RestClient::Resource.new("http://#{server}:#{port}/redirect").get.request.url
+  end
+
+  # Parses JSON data as returned from Mesos's metrics API
+  # @param data [String] Server response
+  # @return [Integer] Number of failed tasks in Mesos
+  def get_slaves(data)
+    begin
+      slaves = JSON.parse(data)[MesosDiskBalanceCheck.metrics_name]
+    rescue JSON::ParserError
+      raise "Could not parse JSON response: #{data}"
+    end
+
+    if slaves.nil?
+      raise "No metrics for [#{MesosDiskBalanceCheck.metrics_name}] in server response: #{data}"
+    end
+
+    slaves
+  end
+
+  def get_check_diff(slavelist)
+    begin
+      usages = {}
+      check_diff = {}
+      slavelist.each do |slaveinfo|
+        usages.store(slaveinfo['hostname'], slaveinfo['used_resources'][CHECK_TYPE] * 100 / slaveinfo['resources'][CHECK_TYPE])
+      end
+      sorted = usages.sort_by { |_hostname, total| total }
+      max = usages.length - 1
+      check_diff['diff'] = sorted[max][1] - sorted[0][1]
+      check_diff['msg'] = "Hostname #{sorted[0][0]} uses #{sorted[0][1]}% and Hostname #{sorted[max][0]} uses #{sorted[max][1]}%"
+    end
+    check_diff
+  end
+end