Module: God

Defined in:
lib/god.rb,
lib/god/task.rb,
lib/god/watch.rb,
lib/god/driver.rb,
lib/god/errors.rb,
lib/god/logger.rb,
lib/god/metric.rb,
lib/god/socket.rb,
lib/god/cli/run.rb,
lib/god/contact.rb,
lib/god/process.rb,
lib/god/trigger.rb,
lib/god/behavior.rb,
lib/god/registry.rb,
lib/god/timeline.rb,
lib/god/condition.rb,
lib/god/cli/command.rb,
lib/god/cli/version.rb,
lib/god/configurable.rb,
lib/god/event_handler.rb,
lib/god/simple_logger.rb,
lib/god/contacts/email.rb,
lib/god/contacts/prowl.rb,
lib/god/contacts/scout.rb,
lib/god/system/process.rb,
lib/god/contacts/jabber.rb,
lib/god/conditions/tries.rb,
lib/god/contacts/twitter.rb,
lib/god/contacts/webhook.rb,
lib/god/conditions/always.rb,
lib/god/conditions/lambda.rb,
lib/god/contacts/campfire.rb,
lib/god/conditions/complex.rb,
lib/god/conditions/flapping.rb,
lib/god/conditions/cpu_usage.rb,
lib/god/conditions/disk_usage.rb,
lib/god/conditions/file_mtime.rb,
lib/god/system/portable_poller.rb,
lib/god/conditions/memory_usage.rb,
lib/god/behaviors/clean_pid_file.rb,
lib/god/conditions/process_exits.rb,
lib/god/system/slash_proc_poller.rb,
lib/god/conditions/process_running.rb,
lib/god/behaviors/clean_unix_socket.rb,
lib/god/conditions/degrading_lambda.rb,
lib/god/conditions/socket_responding.rb,
lib/god/event_handlers/dummy_handler.rb,
lib/god/conditions/http_response_code.rb,
lib/god/event_handlers/kqueue_handler.rb,
lib/god/behaviors/notify_when_flapping.rb,
lib/god/event_handlers/netlink_handler.rb

Defined Under Namespace

Modules: Behaviors, CLI, Conditions, Configurable, Contacts, System Classes: AbstractMethodNotOverriddenError, Behavior, Condition, Contact, Driver, DriverEvent, DriverEventQueue, DriverOperation, DummyHandler, EventCondition, EventHandler, EventRegistrationFailedError, InvalidCommandError, KQueueHandler, Logger, Metric, NetlinkHandler, NoSuchBehaviorError, NoSuchConditionError, NoSuchContactError, NoSuchWatchError, PollCondition, Process, Registry, SimpleLogger, Socket, Task, TimedEvent, Timeline, Trigger, TriggerCondition, Watch

Constant Summary collapse

VERSION =

The String version number for this package.

'0.13.0'
LOG_BUFFER_SIZE_DEFAULT =

The Integer number of lines of backlog to keep for the logger.

100
PID_FILE_DIRECTORY_DEFAULTS =

An Array of directory paths to be used as the default PID file directory. This list will be searched in order and the first one that has write permissions will be used.

['/var/run/god', '~/.god/pids']
DRB_PORT_DEFAULT =

The default Integer port number for the DRb communcations channel.

17165
DRB_ALLOW_DEFAULT =

The default Array of String IPs that will allow DRb communication access.

['127.0.0.1']
LOG_LEVEL_DEFAULT =

The default Symbol log level.

:info
TERMINATE_TIMEOUT_DEFAULT =

The default Integer number of seconds to wait for god to terminate when issued the quit command.

10
STOP_TIMEOUT_DEFAULT =

The default Integer number of seconds to wait for a process to terminate.

10
STOP_SIGNAL_DEFAULT =

The default String signal to send for the stop command.

'TERM'

Class Attribute Summary collapse

Class Method Summary collapse

Class Attribute Details

.contact_groups ⇒ Object

internal



200
201
202
 
# File 'lib/god.rb', line 200

def contact_groups
  @contact_groups
end

.contacts ⇒ Object

internal



200
201
202
 
# File 'lib/god.rb', line 200

def contacts
  @contacts
end

.groups ⇒ Object

internal



200
201
202
 
# File 'lib/god.rb', line 200

def groups
  @groups
end

.inited ⇒ Object

internal



200
201
202
 
# File 'lib/god.rb', line 200

def inited
  @inited
end

.main ⇒ Object

internal



200
201
202
 
# File 'lib/god.rb', line 200

def main
  @main
end

.pending_watch_states ⇒ Object

internal



200
201
202
 
# File 'lib/god.rb', line 200

def pending_watch_states
  @pending_watch_states
end

.pending_watches ⇒ Object

internal



200
201
202
 
# File 'lib/god.rb', line 200

def pending_watches
  @pending_watches
end

.running ⇒ Object

internal



200
201
202
 
# File 'lib/god.rb', line 200

def running
  @running
end

.server ⇒ Object

internal



200
201
202
 
# File 'lib/god.rb', line 200

def server
  @server
end

.watches ⇒ Object

internal



200
201
202
 
# File 'lib/god.rb', line 200

def watches
  @watches
end

Class Method Details

.at_exit ⇒ Object

To be called on program exit to start god.

Returns nothing.



721
722
723
724
 
# File 'lib/god.rb', line 721

def self.at_exit
  self.start
  self.join
end

.contact(kind) {|c| ... } ⇒ Object

Instantiate a new Contact of the given kind and send it to the block. Then prepare, validate, and record the Contact. Aborts on invalid kind, duplicate contact name, invalid contact, or conflicting group name.

kind - The Symbol contact class specifier.

Returns nothing.

Yields:

  • (c) 


363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
 
# File 'lib/god.rb', line 363

def self.contact(kind)
  # Ensure internal init has run.
  self.internal_init

  # Verify contact has been loaded.
  if CONTACT_LOAD_SUCCESS[kind] == false
    applog(nil, :error, "A required dependency for the #{kind} contact is unavailable.")
    applog(nil, :error, "Run the following commands to install the dependencies:")
    CONTACT_DEPS[kind].each do |d|
      applog(nil, :error, "  [sudo] gem install #{d}")
    end
    abort
  end

  # Create the contact.
  begin
    c = Contact.generate(kind)
  rescue NoSuchContactError => e
    abort e.message
  end

  # Send to block so config can set attributes.
  yield(c) if block_given?

  # Call prepare on the contact.
  c.prepare

  # Remove existing contacts of same name.
  existing_contact = self.contacts[c.name]
  if self.running && existing_contact
    self.uncontact(existing_contact)
  end

  # Warn and noop if the contact has been defined before.
  if self.contacts[c.name] || self.contact_groups[c.name]
    applog(nil, :warn, "Contact name '#{c.name}' already used for a Contact or Contact Group")
    return
  end

  # Abort if the Contact is invalid, the Contact will have printed out its
  # own error messages by now.
  unless Contact.valid?(c) && c.valid?
    abort "Exiting on invalid contact"
  end

  # Add to list of contacts.
  self.contacts[c.name] = c

  # Add to contact group if specified.
  if c.group
    # Ensure group name hasn't been used for a contact already.
    if self.contacts[c.group]
      abort "Contact Group name '#{c.group}' already used for a Contact"
    end

    self.contact_groups[c.group] ||= []
    self.contact_groups[c.group] << c
  end
end

.control(name, command) ⇒ Object

Control the lifecycle of the given task(s).

name - The String name of a task/group. command - The String command to run. Valid commands are:

"start", "monitor", "restart", "stop", "unmonitor", "remove".

Returns an Array of String task names affected by the command.



442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
 
# File 'lib/god.rb', line 442

def self.control(name, command)
  # Get the list of items.
  items = Array(self.watches[name] || self.groups[name]).dup

  jobs = []

  # Do the command.
  case command
    when "start", "monitor"
      items.each { |w| jobs << Thread.new { w.monitor if w.state != :up } }
    when "restart"
      items.each { |w| jobs << Thread.new { w.move(:restart) } }
    when "stop"
      items.each { |w| jobs << Thread.new { w.action(:stop); w.unmonitor if w.state != :unmonitored } }
    when "unmonitor"
      items.each { |w| jobs << Thread.new { w.unmonitor if w.state != :unmonitored } }
    when "remove"
      items.each { |w| self.unwatch(w) }
    else
      raise InvalidCommandError.new
  end

  jobs.each { |j| j.join }

  items.map { |x| x.name }
end

.internal_init ⇒ Object

Initialize internal data.

Returns nothing.



228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
 
# File 'lib/god.rb', line 228

def self.internal_init
  # Only do this once.
  return if self.inited

  # Variable init.
  self.watches = {}
  self.groups = {}
  self.pending_watches = []
  self.pending_watch_states = {}
  self.contacts = {}
  self.contact_groups = {}

  # Set defaults.
  self.log_buffer_size ||= LOG_BUFFER_SIZE_DEFAULT
  self.port ||= DRB_PORT_DEFAULT
  self.allow ||= DRB_ALLOW_DEFAULT
  self.log_level ||= LOG_LEVEL_DEFAULT
  self.terminate_timeout ||= TERMINATE_TIMEOUT_DEFAULT

  # Additional setup.
  self.setup

  # Log level.
  log_level_map = {:debug => Logger::DEBUG,
                   :info => Logger::INFO,
                   :warn => Logger::WARN,
                   :error => Logger::ERROR,
                   :fatal => Logger::FATAL}
  LOG.level = log_level_map[self.log_level]

  # Init has been executed.
  self.inited = true

  # Not yet running.
  self.running = false
end

.join ⇒ Object

Prevent god from exiting.

Returns nothing.



709
710
711
 
# File 'lib/god.rb', line 709

def self.join
  self.main.join if self.main
end

.load(glob) ⇒ Object

Load the given file(s) according to the given glob.

glob - The glob-enabled String path to load.

Returns nothing.



627
628
629
630
631
 
# File 'lib/god.rb', line 627

def self.load(glob)
  Dir[glob].each do |f|
    Kernel.load f
  end
end

.pattern_match(pattern, list) ⇒ Object

Match a shortened pattern against a list of String candidates. The pattern is expanded into a regular expression by inserting .* between each character.

pattern - The String containing the abbreviation. list - The Array of Strings to match against.

Examples

list = %w{ foo bar bars }
pattern = 'br'
God.pattern_match(list, pattern)
# => ['bar', 'bars']

Returns the Array of matching name Strings.



743
744
745
746
747
748
749
 
# File 'lib/god.rb', line 743

def self.pattern_match(pattern, list)
  regex = pattern.split('').join('.*')

  list.select do |item|
    item =~ Regexp.new(regex)
  end.sort_by { |x| x.size }
end

.registry ⇒ Object 



2
3
4
 
# File 'lib/god/registry.rb', line 2

def self.registry
  @registry ||= Registry.new
end

.running_load(code, filename, action = nil) ⇒ Object

Load a config file into a running god instance. Rescues any exceptions that the config may raise and reports these back to the caller.

code - The String config file contents. filename - The filename of the config file. action - The optional String command specifying how to deal with

existing watches. Valid options are: 'stop', 'remove' or
'leave' (default).

Returns a three-tuple Array [loaded_names, errors, unloaded_names] where:

loaded_names   - The Array of String task names that were loaded.
errors         - The String of error messages produced during the
                 load phase. Will be a blank String if no errors
                 were encountered.
unloaded_names - The Array of String task names that were unloaded
                 from the system (if 'remove' or 'stop' was
                 specified as the action).



566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
 
# File 'lib/god.rb', line 566

def self.running_load(code, filename, action = nil)
  errors = ""
  loaded_watches = []
  unloaded_watches = []
  jobs = []

  begin
    LOG.start_capture

    Gem.clear_paths
    eval(code, root_binding, filename)
    self.pending_watches.each do |w|
      if previous_state = self.pending_watch_states[w.name]
        w.monitor unless previous_state == :unmonitored
      else
        w.monitor if w.autostart?
      end
    end
    loaded_watches = self.pending_watches.map { |w| w.name }
    self.pending_watches.clear
    self.pending_watch_states.clear

    self.watches.each do |name, watch|
      next if loaded_watches.include?(name)

      case action
      when 'stop'
        jobs << Thread.new(watch) { |w| w.action(:stop); self.unwatch(w) }
        unloaded_watches << name
      when 'remove'
        jobs << Thread.new(watch) { |w| self.unwatch(w) }
        unloaded_watches << name
      when 'leave', '', nil
        # Do nothing
      else
        raise InvalidCommandError, "Unknown action: #{action}"
      end
    end

    # Make sure we quit capturing when we're done.
    LOG.finish_capture
  rescue Exception => e
    # Don't ever let running_load take down god.
    errors << LOG.finish_capture

    unless e.instance_of?(SystemExit)
      errors << e.message << "\n"
      errors << e.backtrace.join("\n")
    end
  end

  jobs.each { |t| t.join }

  [loaded_watches, errors, unloaded_watches]
end

.running_log(watch_name, since) ⇒ Object

Log lines for the given task since the specified time.

watch_name - The String name of the task (may be abbreviated). since - The Time since which to report log lines.

Raises God::NoSuchWatchError if no tasks matched. Returns the String of newline separated log lines.



539
540
541
542
543
544
545
546
547
 
# File 'lib/god.rb', line 539

def self.running_log(watch_name, since)
  matches = pattern_match(watch_name, self.watches.keys)

  unless matches.first
    raise NoSuchWatchError.new
  end

  LOG.watch_log_since(matches.first, since)
end

.setup ⇒ Object

Setup pid file directory and log system.

Returns nothing.



636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
 
# File 'lib/god.rb', line 636

def self.setup
  if self.pid_file_directory
    # Pid file dir was specified, ensure it is created and writable.
    unless File.exist?(self.pid_file_directory)
      begin
        FileUtils.mkdir_p(self.pid_file_directory)
      rescue Errno::EACCES => e
        abort "Failed to create pid file directory: #{e.message}"
      end
    end

    unless File.writable?(self.pid_file_directory)
      abort "The pid file directory (#{self.pid_file_directory}) is not writable by #{Etc.getlogin}"
    end
  else
    # No pid file dir specified, try defaults.
    PID_FILE_DIRECTORY_DEFAULTS.each do |idir|
      dir = File.expand_path(idir)
      begin
        FileUtils.mkdir_p(dir)
        if File.writable?(dir)
          self.pid_file_directory = dir
          break
        end
      rescue Errno::EACCES => e
      end
    end

    unless self.pid_file_directory
      dirs = PID_FILE_DIRECTORY_DEFAULTS.map { |x| File.expand_path(x) }
      abort "No pid file directory exists, could be created, or is writable at any of #{dirs.join(', ')}"
    end
  end

  if God::Logger.syslog
    LOG.info("Syslog enabled.")
  else
    LOG.info("Syslog disabled.")
  end

  applog(nil, :info, "Using pid file directory: #{self.pid_file_directory}")
end

.signal(name, signal) ⇒ Object

Send a signal to each task.

name - The String name of the task or group. signal - The String or integer signal to send. e.g. ‘HUP’, 9.

Returns an Array of String names of the tasks affected.



524
525
526
527
528
529
530
 
# File 'lib/god.rb', line 524

def self.signal(name, signal)
  items = Array(self.watches[name] || self.groups[name]).dup
  jobs = []
  items.each { |w| jobs << Thread.new { w.signal(signal) } }
  jobs.each { |j| j.join }
  items.map { |x| x.name }
end

.start ⇒ Object

Initialize and startup the machinery that makes god work.

Returns nothing.



682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
 
# File 'lib/god.rb', line 682

def self.start
  self.internal_init

  # Instantiate server.
  self.server = Socket.new(self.port, self.socket_user, self.socket_group, self.socket_perms)

  # Start monitoring any watches set to autostart.
  self.watches.values.each { |w| w.monitor if w.autostart? }

  # Clear pending watches.
  self.pending_watches.clear

  # Mark as running.
  self.running = true

  # Don't exit.
  self.main =
  Thread.new do
    loop do
      sleep 60
    end
  end
end

.status ⇒ Object

Gather the status of each task.

Examples

God.status
# => { 'mongrel' => :up, 'nginx' => :up }

Returns a Hash where the key is the String task name and the value is the

Symbol status.



510
511
512
513
514
515
516
 
# File 'lib/god.rb', line 510

def self.status
  info = {}
  self.watches.map do |name, w|
    info[name] = {:state => w.state, :group => w.group}
  end
  info
end

.stop_all ⇒ Object

Unmonitor and stop all tasks.

Returns true on success, false if all tasks could not be stopped within 10 seconds



473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
 
# File 'lib/god.rb', line 473

def self.stop_all
  self.watches.sort.each do |name, w|
    Thread.new do
      w.unmonitor if w.state != :unmonitored
      w.action(:stop) if w.alive?
    end
  end

  terminate_timeout.times do
    return true unless self.watches.map { |name, w| w.alive? }.any?
    sleep 1
  end

  return false
end

.task(klass = Task) {|t| ... } ⇒ Object

Instantiate a new, empty Task object and yield it to the mandatory block. The attributes of the task will be set by the configuration file. Aborts on duplicate task name, invalid task, or conflicting group name.

Returns nothing.

Yields:

  • (t) 


279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
 
# File 'lib/god.rb', line 279

def self.task(klass = Task)
  # Ensure internal init has run.
  self.internal_init

  t = klass.new
  yield(t)

  # Do the post-configuration.
  t.prepare

  # If running, completely remove the watch (if necessary) to prepare for
  # the reload
  existing_watch = self.watches[t.name]
  if self.running && existing_watch
    self.pending_watch_states[existing_watch.name] = existing_watch.state
    self.unwatch(existing_watch)
  end

  # Ensure the new watch has a unique name.
  if self.watches[t.name] || self.groups[t.name]
    abort "Task name '#{t.name}' already used for a Task or Group"
  end

  # Ensure watch is internally valid.
  t.valid? || abort("Task '#{t.name}' is not valid (see above)")

  # Add to list of watches.
  self.watches[t.name] = t

  # Add to pending watches.
  self.pending_watches << t

  # Add to group if specified.
  if t.group
    # Ensure group name hasn't been used for a watch already.
    if self.watches[t.group]
      abort "Group name '#{t.group}' already used for a Task"
    end

    self.groups[t.group] ||= []
    self.groups[t.group] << t
  end

  # Register watch.
  t.register!

  # Log.
  if self.running && existing_watch
    applog(t, :info, "#{t.name} Reloaded config")
  elsif self.running
    applog(t, :info, "#{t.name} Loaded config")
  end
end

.terminate ⇒ Object

Force the termination of god.

  • Clean up pid file if one exists

  • Stop DRb service

  • Hard exit using exit!

Never returns because the process will no longer exist!



495
496
497
498
499
 
# File 'lib/god.rb', line 495

def self.terminate
  FileUtils.rm_f(self.pid) if self.pid
  self.server.stop if self.server
  exit!(0)
end

.uncontact(contact) ⇒ Object

Remove the given contact from god.

contact - The Contact to remove.

Returns nothing.



428
429
430
431
432
433
 
# File 'lib/god.rb', line 428

def self.uncontact(contact)
  self.contacts.delete(contact.name)
  if contact.group
    self.contact_groups[contact.group].delete(contact)
  end
end

.unwatch(watch) ⇒ Object

Unmonitor and remove the given watch from god.

watch - The Watch to remove.

Returns nothing.



338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
 
# File 'lib/god.rb', line 338

def self.unwatch(watch)
  # Unmonitor.
  watch.unmonitor unless watch.state == :unmonitored

  # Unregister.
  watch.unregister!

  # Remove from watches.
  self.watches.delete(watch.name)

  # Remove from groups.
  if watch.group
    self.groups[watch.group].delete(watch)
  end

  applog(watch, :info, "#{watch.name} unwatched")
end

.version ⇒ Object

Returns the version String.



714
715
716
 
# File 'lib/god.rb', line 714

def self.version
  God::VERSION
end

.watch(&block) ⇒ Object

Instantiate a new, empty Watch object and pass it to the mandatory block. The attributes of the watch will be set by the configuration file. Aborts on duplicate watch name, invalid watch, or conflicting group name.

Returns nothing.



270
271
272
 
# File 'lib/god.rb', line 270

def self.watch(&block)
  self.task(Watch, &block)
end