Start of work folding threading and forking togetherthreading_plus_forking

3 files changed, 325 insertions, 7 deletions

diff --git a/lib/ansible/executor/process/threading.py b/lib/ansible/executor/process/threading.py
new file mode 100644
index 0000000000..0f249b4609
--- /dev/null
+++ b/lib/ansible/executor/process/threading.py
@@ -0,0 +1,116 @@
+# (c) 2012-2014, Michael DeHaan <michael.dehaan@gmail.com>
+#
+# This file is part of Ansible
+#
+# Ansible is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# Ansible is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with Ansible.  If not, see <http://www.gnu.org/licenses/>.
+
+# Make coding more python3-ish
+from __future__ import (absolute_import, division, print_function)
+__metaclass__ = type
+
+import os
+import sys
+import time
+import traceback
+
+from jinja2.exceptions import TemplateNotFound
+
+from ansible.errors import AnsibleConnectionFailure
+from ansible.executor.task_executor import TaskExecutor
+from ansible.executor.task_result import TaskResult
+from ansible.module_utils._text import to_text
+
+try:
+    from __main__ import display
+except ImportError:
+    from ansible.utils.display import Display
+    display = Display()
+
+__all__ = ['WorkerProcess']
+
+
+def run_worker(tqm, shared_loader_obj):
+    '''
+    The worker thread class, which uses TaskExecutor to run tasks
+    read from a job queue and pushes results into a results queue
+    for reading later.
+    '''
+
+    # import cProfile, pstats, StringIO
+    # pr = cProfile.Profile()
+    # pr.enable()
+
+    display.debug("STARTING WORKER")
+    while not tqm._terminated:
+        job = tqm.get_job()
+        if job is None:
+            time.sleep(0.0001)
+            continue
+
+        display.debug("WORKER GOT A JOB")
+        (host, task, play_context, task_vars) = job
+
+        try:
+            # execute the task and build a TaskResult from the result
+            display.debug("running TaskExecutor() for %s/%s" % (host, task))
+            executor_result = TaskExecutor(
+                host,
+                task,
+                task_vars,
+                play_context,
+                None, #new_stdin
+                tqm._loader,
+                shared_loader_obj,
+                tqm, #rslt_q
+            ).run()
+
+            display.debug("done running TaskExecutor() for %s/%s" % (host, task))
+
+            # put the result on the result queue
+            display.debug("sending task result")
+            tqm.put_result(TaskResult(
+                host,
+                task,
+                executor_result,
+            ))
+            display.debug("done task result")
+
+        except AnsibleConnectionFailure:
+            tqm.put_result(TaskResult(
+                host,
+                task,
+                dict(unreachable=True),
+            ))
+
+        except Exception as e:
+            if not isinstance(e, (IOError, EOFError, KeyboardInterrupt, SystemExit)) or isinstance(e, TemplateNotFound):
+                try:
+                    tqm.put_result(TaskResult(
+                        host,
+                        task,
+                        dict(failed=True, exception=to_text(traceback.format_exc()), stdout=''),
+                    ))
+                except:
+                    display.debug(u"WORKER EXCEPTION: %s" % to_text(e))
+                    display.debug(u"WORKER TRACEBACK: %s" % to_text(traceback.format_exc()))
+
+    # pr.disable()
+    # s = StringIO.StringIO()
+    # sortby = 'time'
+    # ps = pstats.Stats(pr, stream=s).sort_stats(sortby)
+    # ps.print_stats()
+    # with open('worker_%06d.stats' % os.getpid(), 'w') as f:
+    #     f.write(s.getvalue())
+
+    display.debug("WORKER PROCESS EXITING")
diff --git a/lib/ansible/executor/shared_plugin_loader.py b/lib/ansible/executor/shared_plugin_loader.py
new file mode 100644
index 0000000000..1c3f9fcad3
--- /dev/null
+++ b/lib/ansible/executor/shared_plugin_loader.py
@@ -0,0 +1,35 @@
+# (c) 2017, Red Hat, Inc. <support@ansible.com>
+#
+# This file is part of Ansible
+#
+# Ansible is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# Ansible is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with Ansible.  If not, see <http://www.gnu.org/licenses/>.
+
+# Make coding more python3-ish
+from __future__ import (absolute_import, division, print_function)
+
+from ansible.plugins.loader import action_loader, connection_loader, filter_loader, lookup_loader, module_loader, test_loader
+
+class SharedPluginLoaderObj:
+    '''
+    A simple object to make pass the various plugin loaders to
+    the forked processes over the queue easier
+    '''
+    def __init__(self):
+        self.action_loader = action_loader
+        self.connection_loader = connection_loader
+        self.filter_loader = filter_loader
+        self.lookup_loader = lookup_loader
+        self.module_loader = module_loader
+        self.test_loader = test_loader
+
diff --git a/lib/ansible/executor/task_queue_manager.py b/lib/ansible/executor/task_queue_manager.py
index a70a3923c9..2f9fcab250 100644
--- a/lib/ansible/executor/task_queue_manager.py
+++ b/lib/ansible/executor/task_queue_manager.py
@@ -22,12 +22,20 @@ __metaclass__ = type
 import multiprocessing
 import os
 import tempfile
+import threading
+import time
+
+from collections import deque
 
 from ansible import constants as C
 from ansible.errors import AnsibleError
 from ansible.executor.play_iterator import PlayIterator
+from ansible.executor.process.worker import WorkerProcess
+from ansible.executor.process.threading import run_worker
+from ansible.executor.shared_plugin_loader import SharedPluginLoaderObj
 from ansible.executor.stats import AggregateStats
 from ansible.module_utils.six import string_types
+from ansible.module_utils.six.moves import queue as Queue
 from ansible.module_utils._text import to_text
 from ansible.playbook.block import Block
 from ansible.playbook.play_context import PlayContext
@@ -47,6 +55,26 @@ except ImportError:
 
 __all__ = ['TaskQueueManager']
 
+class ResultsSentinel:
+    pass
+
+_sentinel = ResultsSentinel()
+
+def results_thread_main(tqm):
+    while True:
+        try:
+            result = tqm._final_q.get()
+            if isinstance(result, ResultsSentinel):
+                break
+            else:
+                tqm._results_lock.acquire()
+                tqm._res_queue.append(result)
+                tqm._results_lock.release()
+        except (IOError, EOFError):
+            break
+        except Queue.Empty:
+            pass
+
 
 class TaskQueueManager:
 
@@ -100,18 +128,72 @@ class TaskQueueManager:
         self._failed_hosts = dict()
         self._unreachable_hosts = dict()
 
-        self._final_q = multiprocessing.Queue()
+        # the final results queue we'll use for both forking and threading
+        self._res_queue = deque()
+        self._res_queue_lock = threading.Lock()
+        self._res_ready = threading.Event()
 
         # A temporary file (opened pre-fork) used by connection
         # plugins for inter-process locking.
         self._connection_lockfile = tempfile.TemporaryFile()
 
     def _initialize_processes(self, num):
+        if C.ANSIBLE_PROCESS_MODEL == 'forking':
+            self._final_q = multiprocessing.Queue()
+            self._job_queue = None
+            self._job_queue_lock = None
+            self.put_job = self._forked_put_job
+            self.get_job = self._forked_get_job
+            self.put_result = self._forked_put_result
+            self.get_result = self._forked_get_result
+            self._cleanup_processes = self._cleanup_forked_processes
+            self._initialize_forked_processes(num)
+
+            # create the result processing thread for reading results in the background
+            self._results_lock = threading.Condition(threading.Lock())
+            self._results_thread = threading.Thread(target=results_thread_main, args=(self,))
+            self._results_thread.daemon = True
+            self._results_thread.start()
+
+        elif C.ANSIBLE_PROCESS_MODEL == 'threading':
+            self._job_queue = deque()
+            self._job_queue_lock = threading.Lock()
+            self.put_job = self._threaded_put_job
+            self.get_job = self._threaded_get_job
+            self.put_result = self._threaded_put_result
+            self.get_result = self._threaded_get_result
+            self._cleanup_processes = self._cleanup_threaded_processes
+            self._initialize_threaded_processes(num)
+
+        else:
+            self._cleanup_processes = self._cleanup_dummy
+            raise AnsibleError(
+                      'Invalid process model specified: "%s". ' \
+                      'The process model must be set to either "forking" or "threading"'
+                  )
+
+    def _initialize_forked_processes(self, num):
+        self._workers = []
+        self._cur_worker = 0
+
+        for i in range(num):
+            self._workers.append([None, None])
+
+    def _initialize_threaded_processes(self, num):
+        # FIXME: do we need a global lock for workers here instead of a per-worker?
         self._workers = []
 
+        # create a dummy object with plugin loaders set as an easier
+        # way to share them with the forked processes
+        shared_loader_obj = SharedPluginLoaderObj()
+
         for i in range(num):
             rslt_q = multiprocessing.Queue()
             self._workers.append([None, rslt_q])
+            w_thread = threading.Thread(target=run_worker, args=(self, shared_loader_obj))
+            w_thread.start()
+            w_lock = threading.Lock()
+            self._workers.append([w_thread, w_lock])
 
     def _initialize_notified_handlers(self, play):
         '''
@@ -305,26 +387,36 @@ class TaskQueueManager:
         for host_name in iterator.get_failed_hosts():
             self._failed_hosts[host_name] = True
 
-        strategy.cleanup()
         self._cleanup_processes()
         return play_return
 
     def cleanup(self):
         display.debug("RUNNING CLEANUP")
         self.terminate()
-        self._final_q.close()
+        if hasattr(self, '_final_q'):
+            self._final_q.put(_sentinel)
+            self._results_thread.join()
+            self._final_q.close()
         self._cleanup_processes()
 
-    def _cleanup_processes(self):
+    def _cleanup_dummy(self):
+        return
+
+    def _cleanup_forked_processes(self):
         if hasattr(self, '_workers'):
-            for (worker_prc, rslt_q) in self._workers:
-                rslt_q.close()
+            for (worker_prc, _) in self._workers:
                 if worker_prc and worker_prc.is_alive():
                     try:
                         worker_prc.terminate()
                     except AttributeError:
                         pass
 
+    def _cleanup_threaded_processes(self):
+        if hasattr(self, '_workers'):
+            for (w_thread, w_lock) in self._workers:
+                if w_thread and not w_thread.is_alive():
+                    w_thread.join()
+
     def clear_failed_hosts(self):
         self._failed_hosts = dict()
 
@@ -338,7 +430,7 @@ class TaskQueueManager:
         return self._loader
 
     def get_workers(self):
-        return self._workers[:]
+        return self._workers
 
     def terminate(self):
         self._terminated = True
@@ -380,3 +472,78 @@ class TaskQueueManager:
                     from traceback import format_tb
                     from sys import exc_info
                     display.vvv('Callback Exception: \n' + ' '.join(format_tb(exc_info()[2])))
+
+    # helpers for forking
+    def _forked_put_job(self, data):
+        try:
+            (host, task, play_context, task_vars) = data
+
+            # create a dummy object with plugin loaders set as an easier
+            # way to share them with the forked processes
+            shared_loader_obj = SharedPluginLoaderObj()
+
+            queued = False
+            starting_worker = self._cur_worker
+            while True:
+                (worker_prc, rslt_q) = self._workers[self._cur_worker]
+                if worker_prc is None or not worker_prc.is_alive():
+                    worker_prc = WorkerProcess(self._final_q, task_vars, host, task, play_context, self._loader, self._variable_manager, shared_loader_obj)
+                    self._workers[self._cur_worker][0] = worker_prc
+                    worker_prc.start()
+                    display.debug("worker is %d (out of %d available)" % (self._cur_worker + 1, len(self._workers)))
+                    queued = True
+                self._cur_worker += 1
+                if self._cur_worker >= len(self._workers):
+                    self._cur_worker = 0
+                if queued:
+                    break
+                elif self._cur_worker == starting_worker:
+                    time.sleep(0.0001)
+
+            return True
+        except (EOFError, IOError, AssertionError) as e:
+            # most likely an abort
+            display.debug("got an error while queuing: %s" % e)
+            return False
+
+    def _forked_get_job(self):
+        pass
+
+    def _forked_put_result(self):
+        pass
+
+    def _forked_get_result(self):
+        return self._pop_off_queue(self._res_queue, self._res_queue_lock)
+
+    # helpers for threading
+    def _put_in_queue(self, data, queue, lock):
+        lock.acquire()
+        queue.appendleft(data)
+        lock.release()
+
+    def _pop_off_queue(self, queue, lock):
+        try:
+            data = None
+            lock.acquire()
+            data = queue.pop()
+        except:
+            pass
+        finally:
+            lock.release()
+        return data
+
+    def _threaded_put_job(self, data):
+        self._put_in_queue(data, self._job_queue, self._job_queue_lock)
+        return True
+
+    def _threaded_get_job(self):
+        return self._pop_off_queue(self._job_queue, self._job_queue_lock)
+
+    def _threaded_put_result(self, data):
+        self._put_in_queue(data, self._res_queue, self._res_queue_lock)
+        self._res_ready.set()
+        return True
+
+    def _threaded_get_result(self):
+        return self._pop_off_queue(self._res_queue, self._res_queue_lock)
+