Class: OpenC3::Operator

Inherits:
Object show all
Defined in:
lib/openc3/operators/operator.rb

Direct Known Subclasses

MicroserviceOperator

Constant Summary collapse

CYCLE_TIME =

cycle time to check for new microservices

5.0
PROCESS_SHUTDOWN_SECONDS =
5.0
@@instance =
nil

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initializeOperator

Returns a new instance of Operator.



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
# File 'lib/openc3/operators/operator.rb', line 235

def initialize
  Logger.level = Logger::INFO
  Logger.microservice_name = 'MicroserviceOperator'

  OperatorProcess.setup()
  @cycle_time = (ENV['OPERATOR_CYCLE_TIME'] and ENV['OPERATOR_CYCLE_TIME'].to_f) || CYCLE_TIME # time in seconds
  # Maximum number of new microservices to start per cycle. This spreads a
  # large startup burst (e.g. installing a plugin with many targets) across
  # multiple cycles so the shared services (object store, redis) aren't
  # stampeded by every microservice connecting at once. 0 = no limit.
  @max_start_per_cycle = (ENV['OPENC3_OPERATOR_MAX_START_PER_CYCLE'] || 5).to_i

  @ruby_process_name = ENV['OPENC3_RUBY']
  if RUBY_ENGINE != 'ruby'
    @ruby_process_name ||= 'jruby'
  else
    @ruby_process_name ||= 'ruby'
  end

  @processes = {}
  @new_processes = {}
  @changed_processes = {}
  @removed_processes = {}
  @mutex = Mutex.new
  @shutdown = false
  @shutdown_complete = false
end

Instance Attribute Details

#cycle_timeObject (readonly)

Returns the value of attribute cycle_time.



228
229
230
# File 'lib/openc3/operators/operator.rb', line 228

def cycle_time
  @cycle_time
end

#processesObject (readonly)

Returns the value of attribute processes.



228
229
230
# File 'lib/openc3/operators/operator.rb', line 228

def processes
  @processes
end

Class Method Details

.instanceObject



416
417
418
# File 'lib/openc3/operators/operator.rb', line 416

def self.instance
  @@instance
end

.processesObject



412
413
414
# File 'lib/openc3/operators/operator.rb', line 412

def self.processes
  @@instance.processes
end

.runObject



407
408
409
410
# File 'lib/openc3/operators/operator.rb', line 407

def self.run
  @@instance = self.new
  @@instance.run
end

Instance Method Details

#remove_oldObject



308
309
310
311
312
313
314
315
316
# File 'lib/openc3/operators/operator.rb', line 308

def remove_old
  @mutex.synchronize do
    if @removed_processes.length > 0
      Logger.info("Shutting down #{@removed_processes.length} removed microservices...")
      shutdown_processes(@removed_processes)
      @removed_processes = {}
    end
  end
end

#respawn_changedObject



285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
# File 'lib/openc3/operators/operator.rb', line 285

def respawn_changed
  @mutex.synchronize do
    if @changed_processes.length > 0
      # Cycle at most @max_start_per_cycle changed microservices this cycle;
      # any remaining stay queued in @changed_processes and are cycled on
      # later cycles. This avoids a restart stampede when many microservices
      # change at once (e.g. a configmap change) overwhelming shared
      # services. Processes not yet cycled keep running until their turn.
      cycle_names = @changed_processes.keys
      cycle_names = cycle_names[0...@max_start_per_cycle] if @max_start_per_cycle > 0
      cycle = @changed_processes.slice(*cycle_names)
      Logger.info("Cycling #{cycle.length} of #{@changed_processes.length} changed microservices...")
      shutdown_processes(cycle)
      break if @shutdown

      cycle_names.each do |name|
        @changed_processes[name].start
        @changed_processes.delete(name)
      end
    end
  end
end

#respawn_deadObject



318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
# File 'lib/openc3/operators/operator.rb', line 318

def respawn_dead
  @mutex.synchronize do
    @processes.each do |name, p|
      break if @shutdown
      # Skip processes still queued by the per-cycle start limit; they
      # haven't been started yet so they aren't "dead" to be respawned.
      next if @new_processes[name]
      p.output_increment
      unless p.alive?
        # Respawn process
        output = p.extract_output
        Logger.error("Unexpected process died... respawning! #{p.cmd_line}\n#{output}\n", scope: p.scope)
        p.hard_stop
        p.start
      end
    end
  end
end

#runObject



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
# File 'lib/openc3/operators/operator.rb', line 377

def run
  # Use at_exit to shutdown cleanly
  at_exit { shutdown() }

  # Monitor processes and respawn if died
  Logger.info("#{self.class} Monitoring processes every #{@cycle_time} sec...")
  loop do
    update()
    remove_old()
    respawn_changed()
    start_new()
    respawn_dead()
    break if @shutdown

    sleep(@cycle_time)
    break if @shutdown
  end

  loop do
    break if @shutdown_complete
    sleep(0.1)
  end
ensure
  Logger.info("#{self.class} shutdown complete")
end

#shutdownObject



367
368
369
370
371
372
373
374
375
# File 'lib/openc3/operators/operator.rb', line 367

def shutdown
  @shutdown = true
  @mutex.synchronize do
    Logger.info("Shutting down processes...")
    shutdown_processes(@processes)
    Logger.info("Shutting down processes complete")
    @shutdown_complete = true
  end
end

#shutdown_processes(processes) ⇒ Object



337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
# File 'lib/openc3/operators/operator.rb', line 337

def shutdown_processes(processes)
  # Make a copy so we don't mutate original
  hard_stop_processes = processes.dup
  processes = processes.dup

  Logger.info("Commanding soft stops...")
  processes.each { |_name, p| p.soft_stop }
  start_time = Time.now
  # Allow sufficient time for processes to shutdown cleanly
  while (Time.now - start_time) < PROCESS_SHUTDOWN_SECONDS
    processes_to_remove = []
    processes.each do |name, p|
      unless p.alive?
        processes_to_remove << name
        Logger.debug("Soft stop process successful: #{p.cmd_line}", scope: p.scope)
      end
    end
    processes_to_remove.each do |name|
      processes.delete(name)
    end
    if processes.length <= 0
      Logger.debug("Soft stop all successful")
      break
    end
    sleep(0.1)
  end
  Logger.debug("Commanding hard stops...")
  hard_stop_processes.each { |_name, p| p.output_increment; p.extract_output; p.hard_stop }
end

#start_newObject



267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
# File 'lib/openc3/operators/operator.rb', line 267

def start_new
  @mutex.synchronize do
    if @new_processes.length > 0
      # Start at most @max_start_per_cycle processes this cycle; any
      # remaining stay queued in @new_processes and start on later cycles.
      # This avoids a startup stampede when many microservices appear at
      # once (e.g. a plugin install) overwhelming the object store / redis.
      start_names = @new_processes.keys
      start_names = start_names[0...@max_start_per_cycle] if @max_start_per_cycle > 0
      Logger.info("#{self.class} starting #{start_names.length} of #{@new_processes.length} new process(es)...")
      start_names.each do |name|
        @new_processes[name].start
        @new_processes.delete(name)
      end
    end
  end
end

#stopObject



403
404
405
# File 'lib/openc3/operators/operator.rb', line 403

def stop
  @shutdown = true
end

#updateObject



263
264
265
# File 'lib/openc3/operators/operator.rb', line 263

def update
  raise "Implement in subclass"
end