Performance improvements

* Speed up serialization of hostvars by simply using the internal
  dictionary used for cached lookups
* Use blocking gets/puts on queues instead of spin locking
* Merge sequential implicitly created blocks
This commit is contained in:
James Cammarata 2015-08-17 13:45:07 -04:00
parent 1d402ae390
commit e7b2308b66
5 changed files with 57 additions and 42 deletions

View file

@ -72,9 +72,9 @@ class ResultProcess(multiprocessing.Process):
self._cur_worker = 0
try:
if not rslt_q.empty():
if rslt_q.qsize() > 0:
debug("worker %d has data to read" % self._cur_worker)
result = rslt_q.get(block=False)
result = rslt_q.get()
debug("got a result from worker %d: %s" % (self._cur_worker, result))
break
except queue.Empty:
@ -102,7 +102,7 @@ class ResultProcess(multiprocessing.Process):
try:
result = self._read_worker_result()
if result is None:
time.sleep(0.1)
time.sleep(0.01)
continue
# if this task is registering a result, do it now

View file

@ -92,40 +92,36 @@ class WorkerProcess(multiprocessing.Process):
while True:
task = None
try:
if not self._main_q.empty():
debug("there's work to be done!")
(host, task, basedir, job_vars, play_context, shared_loader_obj) = self._main_q.get(block=False)
debug("got a task/handler to work on: %s" % task)
(host, task, basedir, job_vars, play_context, shared_loader_obj) = self._main_q.get()
debug("there's work to be done!")
debug("got a task/handler to work on: %s" % task)
# because the task queue manager starts workers (forks) before the
# playbook is loaded, set the basedir of the loader inherted by
# this fork now so that we can find files correctly
self._loader.set_basedir(basedir)
# because the task queue manager starts workers (forks) before the
# playbook is loaded, set the basedir of the loader inherted by
# this fork now so that we can find files correctly
self._loader.set_basedir(basedir)
# Serializing/deserializing tasks does not preserve the loader attribute,
# since it is passed to the worker during the forking of the process and
# would be wasteful to serialize. So we set it here on the task now, and
# the task handles updating parent/child objects as needed.
task.set_loader(self._loader)
# Serializing/deserializing tasks does not preserve the loader attribute,
# since it is passed to the worker during the forking of the process and
# would be wasteful to serialize. So we set it here on the task now, and
# the task handles updating parent/child objects as needed.
task.set_loader(self._loader)
# apply the given task's information to the connection info,
# which may override some fields already set by the play or
# the options specified on the command line
new_play_context = play_context.set_task_and_variable_override(task=task, variables=job_vars)
# apply the given task's information to the connection info,
# which may override some fields already set by the play or
# the options specified on the command line
new_play_context = play_context.set_task_and_variable_override(task=task, variables=job_vars)
# execute the task and build a TaskResult from the result
debug("running TaskExecutor() for %s/%s" % (host, task))
executor_result = TaskExecutor(host, task, job_vars, new_play_context, self._new_stdin, self._loader, shared_loader_obj).run()
debug("done running TaskExecutor() for %s/%s" % (host, task))
task_result = TaskResult(host, task, executor_result)
# execute the task and build a TaskResult from the result
debug("running TaskExecutor() for %s/%s" % (host, task))
executor_result = TaskExecutor(host, task, job_vars, new_play_context, self._new_stdin, self._loader, shared_loader_obj).run()
debug("done running TaskExecutor() for %s/%s" % (host, task))
task_result = TaskResult(host, task, executor_result)
# put the result on the result queue
debug("sending task result")
self._rslt_q.put(task_result, block=False)
debug("done sending task result")
else:
time.sleep(0.1)
# put the result on the result queue
debug("sending task result")
self._rslt_q.put(task_result)
debug("done sending task result")
except queue.Empty:
pass

View file

@ -37,12 +37,13 @@ class Block(Base, Become, Conditional, Taggable):
# similar to the 'else' clause for exceptions
#_otherwise = FieldAttribute(isa='list')
def __init__(self, play=None, parent_block=None, role=None, task_include=None, use_handlers=False):
def __init__(self, play=None, parent_block=None, role=None, task_include=None, use_handlers=False, implicit=False):
self._play = play
self._role = role
self._task_include = task_include
self._parent_block = parent_block
self._use_handlers = use_handlers
self._implicit = implicit
self._dep_chain = []
super(Block, self).__init__()
@ -66,22 +67,27 @@ class Block(Base, Become, Conditional, Taggable):
@staticmethod
def load(data, play=None, parent_block=None, role=None, task_include=None, use_handlers=False, variable_manager=None, loader=None):
b = Block(play=play, parent_block=parent_block, role=role, task_include=task_include, use_handlers=use_handlers)
implicit = not Block.is_block(data)
b = Block(play=play, parent_block=parent_block, role=role, task_include=task_include, use_handlers=use_handlers, implicit=implicit)
return b.load_data(data, variable_manager=variable_manager, loader=loader)
@staticmethod
def is_block(ds):
is_block = False
if isinstance(ds, dict):
for attr in ('block', 'rescue', 'always'):
if attr in ds:
is_block = True
break
return is_block
def preprocess_data(self, ds):
'''
If a simple task is given, an implicit block for that single task
is created, which goes in the main portion of the block
'''
is_block = False
for attr in ('block', 'rescue', 'always'):
if attr in ds:
is_block = True
break
if not is_block:
if not Block.is_block(ds):
if isinstance(ds, list):
return super(Block, self).preprocess_data(dict(block=ds))
else:

View file

@ -52,7 +52,13 @@ def load_list_of_blocks(ds, play, parent_block=None, role=None, task_include=Non
variable_manager=variable_manager,
loader=loader
)
block_list.append(b)
# Implicit blocks are created by bare tasks listed in a play withou
# an explicit block statement. If we have two implicit blocks in a row,
# squash them down to a single block to save processing time later.
if b._implicit and len(block_list) > 0 and block_list[-1]._implicit:
block_list[-1].block.extend(b.block)
else:
block_list.append(b)
return block_list

View file

@ -54,8 +54,15 @@ class HostVars(collections.Mapping):
if item and item is not j2undefined:
return True
return False
def __iter__(self):
raise NotImplementedError('HostVars does not support iteration as hosts are discovered on an as needed basis.')
def __len__(self):
raise NotImplementedError('HostVars does not support len. hosts entries are discovered dynamically as needed')
def __getstate__(self):
return self._lookup
def __setstate__(self, data):
self._lookup = data