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/contacts/sms.rb,
lib/god/event_handler.rb,
lib/god/simple_logger.rb,
lib/god/contacts/email.rb,
lib/god/system/process.rb,
lib/god/conditions/ping.rb,
lib/god/contacts/jabber.rb,
lib/god/conditions/tries.rb,
lib/god/contacts/twitter.rb,
lib/god/contacts/webhook.rb,
lib/god/dependency_graph.rb,
lib/god/dependency_graph.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/condition_helper.rb,
lib/god/conditions/degrading_lambda.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, DependencyGraph, 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

LOG_BUFFER_SIZE_DEFAULT = 
100
PID_FILE_DIRECTORY_DEFAULTS = 
['/var/run/god', '~/.god/pids']
DRB_PORT_DEFAULT = 
17165
DRB_ALLOW_DEFAULT = 
['127.0.0.1']
LOG_LEVEL_DEFAULT = 
:info

Class Attribute Summary collapse

Class Method Summary collapse

Class Attribute Details

.contact_groups ⇒ Object

internal



176
177
178
 
# File 'lib/god.rb', line 176

def contact_groups
  @contact_groups
end

.contacts ⇒ Object

internal



176
177
178
 
# File 'lib/god.rb', line 176

def contacts
  @contacts
end

.groups ⇒ Object

internal



176
177
178
 
# File 'lib/god.rb', line 176

def groups
  @groups
end

.inited ⇒ Object

internal



176
177
178
 
# File 'lib/god.rb', line 176

def inited
  @inited
end

.main ⇒ Object

internal



176
177
178
 
# File 'lib/god.rb', line 176

def main
  @main
end

.pending_watch_states ⇒ Object

internal



176
177
178
 
# File 'lib/god.rb', line 176

def pending_watch_states
  @pending_watch_states
end

.pending_watches ⇒ Object

internal



176
177
178
 
# File 'lib/god.rb', line 176

def pending_watches
  @pending_watches
end

.running ⇒ Object

internal



176
177
178
 
# File 'lib/god.rb', line 176

def running
  @running
end

.server ⇒ Object

internal



176
177
178
 
# File 'lib/god.rb', line 176

def server
  @server
end

.watches ⇒ Object

internal



176
177
178
 
# File 'lib/god.rb', line 176

def watches
  @watches
end

Class Method Details

.at_exit ⇒ Object

To be called on program exit to start god

Returns nothing



641
642
643
 
# File 'lib/god.rb', line 641

def self.at_exit
  self.start
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.

+kind+ is the contact class specifier

Aborts on invalid kind

duplicate contact name
invalid contact
conflicting group name

Returns nothing

Yields:

  • (c) 


343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
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
 
# File 'lib/god.rb', line 343

def self.contact(kind)
  self.internal_init
  
  # 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+ is the name of a task/group (String)
+command+ is the command to run (String)
          one of: "start"
                  "monitor"
                  "restart"
                  "stop"
                  "unmonitor"
                  "remove"

Returns String[]:task_names



414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
 
# File 'lib/god.rb', line 414

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



200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
 
# File 'lib/god.rb', line 200

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
  
  # 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

.load(glob) ⇒ Object

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

+glob+ is the glob-enabled path to load

Returns nothing



561
562
563
564
565
 
# File 'lib/god.rb', line 561

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+ is the String containing the abbreviation
+list+ is the Array of Strings to match against

Examples

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

Returns String[]:matched_elements



661
662
663
664
665
666
667
 
# File 'lib/god.rb', line 661

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) ⇒ 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+ is a String containing the config file
+filename+ is the filename of the config file

Returns [String[]:task_names, String:errors]



524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
 
# File 'lib/god.rb', line 524

def self.running_load(code, filename)
  errors = ""
  watches = []
  
  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
    watches = self.pending_watches.dup
    self.pending_watches.clear
    self.pending_watch_states.clear
  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
  
  names = watches.map { |x| x.name }
  [names, errors]
end

.running_log(watch_name, since) ⇒ Object

Log lines for the given task since the specified time.

+watch_name+ is the name of the task (may be abbreviated)
+since+ is the Time since which to report log lines

Raises God::NoSuchWatchError if no tasks matched

Returns String:joined_log_lines



508
509
510
511
512
513
514
515
516
 
# File 'lib/god.rb', line 508

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 



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
 
# File 'lib/god.rb', line 567

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

.signal(name, signal) ⇒ Object

Send a signal to each task.

+name+ is the String name of the task or group
+signal+ is the signal to send. e.g. HUP, 9

Returns String[]:task_names



493
494
495
496
497
498
499
 
# File 'lib/god.rb', line 493

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



607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
 
# File 'lib/god.rb', line 607

def self.start
  self.internal_init
  
  # instantiate server
  self.server = Socket.new(self.port)
  
  # 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
  
  self.main.join
end

.status ⇒ Object

Gather the status of each task.

Examples

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

Returns { String:task_name => Symbol:status, … }



480
481
482
483
484
485
486
 
# File 'lib/god.rb', line 480

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



445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
 
# File 'lib/god.rb', line 445

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
  
  10.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
conflicting group name

Returns nothing

Yields:

  • (t) 


258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
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
 
# File 'lib/god.rb', line 258

def self.task(klass = Task)
  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!



467
468
469
470
471
 
# File 'lib/god.rb', line 467

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+ is the Contact to remove

Returns nothing



396
397
398
399
400
401
 
# File 'lib/god.rb', line 396

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+ is the Watch to remove

Returns nothing



315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
 
# File 'lib/god.rb', line 315

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 



633
634
635
636
 
# File 'lib/god.rb', line 633

def self.version
  yml = YAML.load(File.read(File.join(File.dirname(__FILE__), *%w[.. VERSION.yml])))
  "#{yml[:major]}.#{yml[:minor]}.#{yml[:patch]}"
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
conflicting group name

Returns nothing



245
246
247
 
# File 'lib/god.rb', line 245

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