Statistics
| Branch: | Tag: | Revision:

root / lib / workerpool.py @ 1a05d855

History | View | Annotate | Download (9 kB)

1 76094e37 Michael Hanselmann
#
2 76094e37 Michael Hanselmann
#
3 76094e37 Michael Hanselmann
4 76094e37 Michael Hanselmann
# Copyright (C) 2008 Google Inc.
5 76094e37 Michael Hanselmann
#
6 76094e37 Michael Hanselmann
# This program is free software; you can redistribute it and/or modify
7 76094e37 Michael Hanselmann
# it under the terms of the GNU General Public License as published by
8 76094e37 Michael Hanselmann
# the Free Software Foundation; either version 2 of the License, or
9 76094e37 Michael Hanselmann
# (at your option) any later version.
10 76094e37 Michael Hanselmann
#
11 76094e37 Michael Hanselmann
# This program is distributed in the hope that it will be useful, but
12 76094e37 Michael Hanselmann
# WITHOUT ANY WARRANTY; without even the implied warranty of
13 76094e37 Michael Hanselmann
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14 76094e37 Michael Hanselmann
# General Public License for more details.
15 76094e37 Michael Hanselmann
#
16 76094e37 Michael Hanselmann
# You should have received a copy of the GNU General Public License
17 76094e37 Michael Hanselmann
# along with this program; if not, write to the Free Software
18 76094e37 Michael Hanselmann
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
19 76094e37 Michael Hanselmann
# 02110-1301, USA.
20 76094e37 Michael Hanselmann
21 76094e37 Michael Hanselmann
22 76094e37 Michael Hanselmann
"""Base classes for worker pools.
23 76094e37 Michael Hanselmann

24 76094e37 Michael Hanselmann
"""
25 76094e37 Michael Hanselmann
26 76094e37 Michael Hanselmann
import collections
27 76094e37 Michael Hanselmann
import logging
28 76094e37 Michael Hanselmann
import threading
29 76094e37 Michael Hanselmann
30 76094e37 Michael Hanselmann
31 76094e37 Michael Hanselmann
class BaseWorker(threading.Thread, object):
32 76094e37 Michael Hanselmann
  """Base worker class for worker pools.
33 76094e37 Michael Hanselmann

34 76094e37 Michael Hanselmann
  Users of a worker pool must override RunTask in a subclass.
35 76094e37 Michael Hanselmann

36 76094e37 Michael Hanselmann
  """
37 76094e37 Michael Hanselmann
  def __init__(self, pool, worker_id):
38 76094e37 Michael Hanselmann
    """Constructor for BaseWorker thread.
39 76094e37 Michael Hanselmann

40 76094e37 Michael Hanselmann
    Args:
41 76094e37 Michael Hanselmann
    - pool: Parent worker pool
42 76094e37 Michael Hanselmann
    - worker_id: Identifier for this worker
43 76094e37 Michael Hanselmann

44 76094e37 Michael Hanselmann
    """
45 76094e37 Michael Hanselmann
    super(BaseWorker, self).__init__()
46 76094e37 Michael Hanselmann
    self.pool = pool
47 76094e37 Michael Hanselmann
    self.worker_id = worker_id
48 76094e37 Michael Hanselmann
    self._current_task = None
49 76094e37 Michael Hanselmann
50 76094e37 Michael Hanselmann
  def ShouldTerminate(self):
51 76094e37 Michael Hanselmann
    """Returns whether a worker should terminate.
52 76094e37 Michael Hanselmann

53 76094e37 Michael Hanselmann
    """
54 76094e37 Michael Hanselmann
    return self.pool.ShouldWorkerTerminate(self)
55 76094e37 Michael Hanselmann
56 b3558df1 Michael Hanselmann
  def _HasRunningTaskUnlocked(self):
57 b3558df1 Michael Hanselmann
    """Returns whether this worker is currently running a task.
58 b3558df1 Michael Hanselmann

59 b3558df1 Michael Hanselmann
    """
60 b3558df1 Michael Hanselmann
    return (self._current_task is not None)
61 b3558df1 Michael Hanselmann
62 b3558df1 Michael Hanselmann
  def HasRunningTask(self):
63 b3558df1 Michael Hanselmann
    """Returns whether this worker is currently running a task.
64 b3558df1 Michael Hanselmann

65 b3558df1 Michael Hanselmann
    """
66 b3558df1 Michael Hanselmann
    self.pool._lock.acquire()
67 b3558df1 Michael Hanselmann
    try:
68 b3558df1 Michael Hanselmann
      return self._HasRunningTaskUnlocked()
69 b3558df1 Michael Hanselmann
    finally:
70 b3558df1 Michael Hanselmann
      self.pool._lock.release()
71 b3558df1 Michael Hanselmann
72 76094e37 Michael Hanselmann
  def run(self):
73 76094e37 Michael Hanselmann
    """Main thread function.
74 76094e37 Michael Hanselmann

75 76094e37 Michael Hanselmann
    Waits for new tasks to show up in the queue.
76 76094e37 Michael Hanselmann

77 76094e37 Michael Hanselmann
    """
78 76094e37 Michael Hanselmann
    pool = self.pool
79 76094e37 Michael Hanselmann
80 b3558df1 Michael Hanselmann
    assert not self.HasRunningTask()
81 76094e37 Michael Hanselmann
82 76094e37 Michael Hanselmann
    while True:
83 76094e37 Michael Hanselmann
      try:
84 76094e37 Michael Hanselmann
        # We wait on lock to be told either terminate or do a task.
85 76094e37 Michael Hanselmann
        pool._lock.acquire()
86 76094e37 Michael Hanselmann
        try:
87 76094e37 Michael Hanselmann
          if pool._ShouldWorkerTerminateUnlocked(self):
88 76094e37 Michael Hanselmann
            break
89 76094e37 Michael Hanselmann
90 76094e37 Michael Hanselmann
          # We only wait if there's no task for us.
91 76094e37 Michael Hanselmann
          if not pool._tasks:
92 b3558df1 Michael Hanselmann
            logging.debug("Worker %s: waiting for tasks", self.worker_id)
93 b3558df1 Michael Hanselmann
94 76094e37 Michael Hanselmann
            # wait() releases the lock and sleeps until notified
95 53b1d12b Michael Hanselmann
            pool._pool_to_worker.wait()
96 76094e37 Michael Hanselmann
97 b3558df1 Michael Hanselmann
            logging.debug("Worker %s: notified while waiting", self.worker_id)
98 b3558df1 Michael Hanselmann
99 76094e37 Michael Hanselmann
            # Were we woken up in order to terminate?
100 76094e37 Michael Hanselmann
            if pool._ShouldWorkerTerminateUnlocked(self):
101 76094e37 Michael Hanselmann
              break
102 76094e37 Michael Hanselmann
103 76094e37 Michael Hanselmann
            if not pool._tasks:
104 76094e37 Michael Hanselmann
              # Spurious notification, ignore
105 76094e37 Michael Hanselmann
              continue
106 76094e37 Michael Hanselmann
107 76094e37 Michael Hanselmann
          # Get task from queue and tell pool about it
108 76094e37 Michael Hanselmann
          try:
109 76094e37 Michael Hanselmann
            self._current_task = pool._tasks.popleft()
110 76094e37 Michael Hanselmann
          finally:
111 53b1d12b Michael Hanselmann
            pool._worker_to_pool.notifyAll()
112 76094e37 Michael Hanselmann
        finally:
113 76094e37 Michael Hanselmann
          pool._lock.release()
114 76094e37 Michael Hanselmann
115 76094e37 Michael Hanselmann
        # Run the actual task
116 76094e37 Michael Hanselmann
        try:
117 b3558df1 Michael Hanselmann
          logging.debug("Worker %s: starting task %r",
118 b3558df1 Michael Hanselmann
                        self.worker_id, self._current_task)
119 76094e37 Michael Hanselmann
          self.RunTask(*self._current_task)
120 b3558df1 Michael Hanselmann
          logging.debug("Worker %s: done with task %r",
121 b3558df1 Michael Hanselmann
                        self.worker_id, self._current_task)
122 76094e37 Michael Hanselmann
        except:
123 76094e37 Michael Hanselmann
          logging.error("Worker %s: Caught unhandled exception",
124 76094e37 Michael Hanselmann
                        self.worker_id, exc_info=True)
125 76094e37 Michael Hanselmann
      finally:
126 76094e37 Michael Hanselmann
        # Notify pool
127 76094e37 Michael Hanselmann
        pool._lock.acquire()
128 76094e37 Michael Hanselmann
        try:
129 b3558df1 Michael Hanselmann
          if self._current_task:
130 b3558df1 Michael Hanselmann
            self._current_task = None
131 53b1d12b Michael Hanselmann
            pool._worker_to_pool.notifyAll()
132 76094e37 Michael Hanselmann
        finally:
133 76094e37 Michael Hanselmann
          pool._lock.release()
134 76094e37 Michael Hanselmann
135 b3558df1 Michael Hanselmann
    logging.debug("Worker %s: terminates", self.worker_id)
136 b3558df1 Michael Hanselmann
137 76094e37 Michael Hanselmann
  def RunTask(self, *args):
138 76094e37 Michael Hanselmann
    """Function called to start a task.
139 76094e37 Michael Hanselmann

140 76094e37 Michael Hanselmann
    """
141 76094e37 Michael Hanselmann
    raise NotImplementedError()
142 76094e37 Michael Hanselmann
143 76094e37 Michael Hanselmann
144 76094e37 Michael Hanselmann
class WorkerPool(object):
145 76094e37 Michael Hanselmann
  """Worker pool with a queue.
146 76094e37 Michael Hanselmann

147 76094e37 Michael Hanselmann
  This class is thread-safe.
148 76094e37 Michael Hanselmann

149 76094e37 Michael Hanselmann
  Tasks are guaranteed to be started in the order in which they're added to the
150 76094e37 Michael Hanselmann
  pool. Due to the nature of threading, they're not guaranteed to finish in the
151 76094e37 Michael Hanselmann
  same order.
152 76094e37 Michael Hanselmann

153 76094e37 Michael Hanselmann
  """
154 76094e37 Michael Hanselmann
  def __init__(self, num_workers, worker_class):
155 76094e37 Michael Hanselmann
    """Constructor for worker pool.
156 76094e37 Michael Hanselmann

157 76094e37 Michael Hanselmann
    Args:
158 76094e37 Michael Hanselmann
    - num_workers: Number of workers to be started (dynamic resizing is not
159 76094e37 Michael Hanselmann
                   yet implemented)
160 76094e37 Michael Hanselmann
    - worker_class: Class to be instantiated for workers; should derive from
161 76094e37 Michael Hanselmann
                    BaseWorker
162 76094e37 Michael Hanselmann

163 76094e37 Michael Hanselmann
    """
164 76094e37 Michael Hanselmann
    # Some of these variables are accessed by BaseWorker
165 53b1d12b Michael Hanselmann
    self._lock = threading.Lock()
166 53b1d12b Michael Hanselmann
    self._pool_to_pool = threading.Condition(self._lock)
167 53b1d12b Michael Hanselmann
    self._pool_to_worker = threading.Condition(self._lock)
168 53b1d12b Michael Hanselmann
    self._worker_to_pool = threading.Condition(self._lock)
169 76094e37 Michael Hanselmann
    self._worker_class = worker_class
170 76094e37 Michael Hanselmann
    self._last_worker_id = 0
171 76094e37 Michael Hanselmann
    self._workers = []
172 76094e37 Michael Hanselmann
    self._quiescing = False
173 76094e37 Michael Hanselmann
174 76094e37 Michael Hanselmann
    # Terminating workers
175 76094e37 Michael Hanselmann
    self._termworkers = []
176 76094e37 Michael Hanselmann
177 76094e37 Michael Hanselmann
    # Queued tasks
178 76094e37 Michael Hanselmann
    self._tasks = collections.deque()
179 76094e37 Michael Hanselmann
180 76094e37 Michael Hanselmann
    # Start workers
181 76094e37 Michael Hanselmann
    self.Resize(num_workers)
182 76094e37 Michael Hanselmann
183 76094e37 Michael Hanselmann
  # TODO: Implement dynamic resizing?
184 76094e37 Michael Hanselmann
185 76094e37 Michael Hanselmann
  def AddTask(self, *args):
186 76094e37 Michael Hanselmann
    """Adds a task to the queue.
187 76094e37 Michael Hanselmann

188 76094e37 Michael Hanselmann
    Args:
189 76094e37 Michael Hanselmann
    - *args: Arguments passed to BaseWorker.RunTask
190 76094e37 Michael Hanselmann

191 76094e37 Michael Hanselmann
    """
192 76094e37 Michael Hanselmann
    self._lock.acquire()
193 76094e37 Michael Hanselmann
    try:
194 76094e37 Michael Hanselmann
      # Don't add new tasks while we're quiescing
195 76094e37 Michael Hanselmann
      while self._quiescing:
196 53b1d12b Michael Hanselmann
        self._pool_to_pool.wait()
197 76094e37 Michael Hanselmann
198 76094e37 Michael Hanselmann
      # Add task to internal queue
199 76094e37 Michael Hanselmann
      self._tasks.append(args)
200 53b1d12b Michael Hanselmann
201 53b1d12b Michael Hanselmann
      # Wake one idling worker up
202 53b1d12b Michael Hanselmann
      self._pool_to_worker.notify()
203 76094e37 Michael Hanselmann
    finally:
204 76094e37 Michael Hanselmann
      self._lock.release()
205 76094e37 Michael Hanselmann
206 76094e37 Michael Hanselmann
  def _ShouldWorkerTerminateUnlocked(self, worker):
207 76094e37 Michael Hanselmann
    """Returns whether a worker should terminate.
208 76094e37 Michael Hanselmann

209 76094e37 Michael Hanselmann
    """
210 76094e37 Michael Hanselmann
    return (worker in self._termworkers)
211 76094e37 Michael Hanselmann
212 76094e37 Michael Hanselmann
  def ShouldWorkerTerminate(self, worker):
213 76094e37 Michael Hanselmann
    """Returns whether a worker should terminate.
214 76094e37 Michael Hanselmann

215 76094e37 Michael Hanselmann
    """
216 76094e37 Michael Hanselmann
    self._lock.acquire()
217 76094e37 Michael Hanselmann
    try:
218 76094e37 Michael Hanselmann
      return self._ShouldWorkerTerminateUnlocked(self)
219 76094e37 Michael Hanselmann
    finally:
220 76094e37 Michael Hanselmann
      self._lock.release()
221 76094e37 Michael Hanselmann
222 76094e37 Michael Hanselmann
  def _HasRunningTasksUnlocked(self):
223 76094e37 Michael Hanselmann
    """Checks whether there's a task running in a worker.
224 76094e37 Michael Hanselmann

225 76094e37 Michael Hanselmann
    """
226 76094e37 Michael Hanselmann
    for worker in self._workers + self._termworkers:
227 b3558df1 Michael Hanselmann
      if worker._HasRunningTaskUnlocked():
228 76094e37 Michael Hanselmann
        return True
229 76094e37 Michael Hanselmann
    return False
230 76094e37 Michael Hanselmann
231 76094e37 Michael Hanselmann
  def Quiesce(self):
232 76094e37 Michael Hanselmann
    """Waits until the task queue is empty.
233 76094e37 Michael Hanselmann

234 76094e37 Michael Hanselmann
    """
235 76094e37 Michael Hanselmann
    self._lock.acquire()
236 76094e37 Michael Hanselmann
    try:
237 76094e37 Michael Hanselmann
      self._quiescing = True
238 76094e37 Michael Hanselmann
239 76094e37 Michael Hanselmann
      # Wait while there are tasks pending or running
240 76094e37 Michael Hanselmann
      while self._tasks or self._HasRunningTasksUnlocked():
241 53b1d12b Michael Hanselmann
        self._worker_to_pool.wait()
242 76094e37 Michael Hanselmann
243 76094e37 Michael Hanselmann
    finally:
244 76094e37 Michael Hanselmann
      self._quiescing = False
245 76094e37 Michael Hanselmann
246 76094e37 Michael Hanselmann
      # Make sure AddTasks continues in case it was waiting
247 53b1d12b Michael Hanselmann
      self._pool_to_pool.notifyAll()
248 76094e37 Michael Hanselmann
249 76094e37 Michael Hanselmann
      self._lock.release()
250 76094e37 Michael Hanselmann
251 76094e37 Michael Hanselmann
  def _NewWorkerIdUnlocked(self):
252 76094e37 Michael Hanselmann
    self._last_worker_id += 1
253 76094e37 Michael Hanselmann
    return self._last_worker_id
254 76094e37 Michael Hanselmann
255 76094e37 Michael Hanselmann
  def _ResizeUnlocked(self, num_workers):
256 76094e37 Michael Hanselmann
    """Changes the number of workers.
257 76094e37 Michael Hanselmann

258 76094e37 Michael Hanselmann
    """
259 76094e37 Michael Hanselmann
    assert num_workers >= 0, "num_workers must be >= 0"
260 76094e37 Michael Hanselmann
261 76094e37 Michael Hanselmann
    logging.debug("Resizing to %s workers", num_workers)
262 76094e37 Michael Hanselmann
263 76094e37 Michael Hanselmann
    current_count = len(self._workers)
264 76094e37 Michael Hanselmann
265 76094e37 Michael Hanselmann
    if current_count == num_workers:
266 76094e37 Michael Hanselmann
      # Nothing to do
267 76094e37 Michael Hanselmann
      pass
268 76094e37 Michael Hanselmann
269 76094e37 Michael Hanselmann
    elif current_count > num_workers:
270 76094e37 Michael Hanselmann
      if num_workers == 0:
271 76094e37 Michael Hanselmann
        # Create copy of list to iterate over while lock isn't held.
272 76094e37 Michael Hanselmann
        termworkers = self._workers[:]
273 76094e37 Michael Hanselmann
        del self._workers[:]
274 76094e37 Michael Hanselmann
      else:
275 76094e37 Michael Hanselmann
        # TODO: Implement partial downsizing
276 76094e37 Michael Hanselmann
        raise NotImplementedError()
277 76094e37 Michael Hanselmann
        #termworkers = ...
278 76094e37 Michael Hanselmann
279 76094e37 Michael Hanselmann
      self._termworkers += termworkers
280 76094e37 Michael Hanselmann
281 76094e37 Michael Hanselmann
      # Notify workers that something has changed
282 53b1d12b Michael Hanselmann
      self._pool_to_worker.notifyAll()
283 76094e37 Michael Hanselmann
284 76094e37 Michael Hanselmann
      # Join all terminating workers
285 76094e37 Michael Hanselmann
      self._lock.release()
286 76094e37 Michael Hanselmann
      try:
287 76094e37 Michael Hanselmann
        for worker in termworkers:
288 c0a8eb9e Michael Hanselmann
          logging.debug("Waiting for thread %s", worker.getName())
289 76094e37 Michael Hanselmann
          worker.join()
290 76094e37 Michael Hanselmann
      finally:
291 76094e37 Michael Hanselmann
        self._lock.acquire()
292 76094e37 Michael Hanselmann
293 76094e37 Michael Hanselmann
      # Remove terminated threads. This could be done in a more efficient way
294 76094e37 Michael Hanselmann
      # (del self._termworkers[:]), but checking worker.isAlive() makes sure we
295 76094e37 Michael Hanselmann
      # don't leave zombie threads around.
296 76094e37 Michael Hanselmann
      for worker in termworkers:
297 76094e37 Michael Hanselmann
        assert worker in self._termworkers, ("Worker not in list of"
298 76094e37 Michael Hanselmann
                                             " terminating workers")
299 76094e37 Michael Hanselmann
        if not worker.isAlive():
300 76094e37 Michael Hanselmann
          self._termworkers.remove(worker)
301 76094e37 Michael Hanselmann
302 76094e37 Michael Hanselmann
      assert not self._termworkers, "Zombie worker detected"
303 76094e37 Michael Hanselmann
304 76094e37 Michael Hanselmann
    elif current_count < num_workers:
305 76094e37 Michael Hanselmann
      # Create (num_workers - current_count) new workers
306 76094e37 Michael Hanselmann
      for i in xrange(num_workers - current_count):
307 76094e37 Michael Hanselmann
        worker = self._worker_class(self, self._NewWorkerIdUnlocked())
308 76094e37 Michael Hanselmann
        self._workers.append(worker)
309 76094e37 Michael Hanselmann
        worker.start()
310 76094e37 Michael Hanselmann
311 76094e37 Michael Hanselmann
  def Resize(self, num_workers):
312 76094e37 Michael Hanselmann
    """Changes the number of workers in the pool.
313 76094e37 Michael Hanselmann

314 76094e37 Michael Hanselmann
    Args:
315 76094e37 Michael Hanselmann
    - num_workers: New number of workers
316 76094e37 Michael Hanselmann

317 76094e37 Michael Hanselmann
    """
318 76094e37 Michael Hanselmann
    self._lock.acquire()
319 76094e37 Michael Hanselmann
    try:
320 76094e37 Michael Hanselmann
      return self._ResizeUnlocked(num_workers)
321 76094e37 Michael Hanselmann
    finally:
322 76094e37 Michael Hanselmann
      self._lock.release()
323 76094e37 Michael Hanselmann
324 76094e37 Michael Hanselmann
  def TerminateWorkers(self):
325 76094e37 Michael Hanselmann
    """Terminate all worker threads.
326 76094e37 Michael Hanselmann

327 76094e37 Michael Hanselmann
    Unstarted tasks will be ignored.
328 76094e37 Michael Hanselmann

329 76094e37 Michael Hanselmann
    """
330 76094e37 Michael Hanselmann
    logging.debug("Terminating all workers")
331 76094e37 Michael Hanselmann
332 76094e37 Michael Hanselmann
    self._lock.acquire()
333 76094e37 Michael Hanselmann
    try:
334 76094e37 Michael Hanselmann
      self._ResizeUnlocked(0)
335 76094e37 Michael Hanselmann
336 76094e37 Michael Hanselmann
      if self._tasks:
337 76094e37 Michael Hanselmann
        logging.debug("There are %s tasks left", len(self._tasks))
338 76094e37 Michael Hanselmann
    finally:
339 76094e37 Michael Hanselmann
      self._lock.release()
340 76094e37 Michael Hanselmann
341 76094e37 Michael Hanselmann
    logging.debug("All workers terminated")