class DparkEnv:
environ = {}
@classmethod
def register(cls, name, value):
cls.environ[name] = value
@classmethod
def get(cls, name, default=None):
return cls.environ.get(name, default)
def __init__(self):
self.started = False
def start(self, isMaster, environ={}, isLocal=False):
if self.started:
return
logger.debug("start env in %s: %s %s", os.getpid(),
isMaster, environ)
self.isMaster = isMaster
self.isLocal = isLocal
if isMaster:
roots = conf.DPARK_WORK_DIR.split(',')
if isLocal:
root = roots[0] # for local mode
if not os.path.exists(root):
os.mkdir(root, 0777)
os.chmod(root, 0777) # because of umask
name = '%s-%s-%d' % (time.strftime("%Y%m%d-%H%M%S"),
socket.gethostname(), os.getpid())
self.workdir = [os.path.join(root, name) for root in roots]
for d in self.workdir:
if not os.path.exists(d):
try: os.makedirs(d)
except OSError: pass
self.environ['WORKDIR'] = self.workdir
self.environ['COMPRESS'] = util.COMPRESS
else:
self.environ.update(environ)
if self.environ['COMPRESS'] != util.COMPRESS:
raise Exception("no %s available" % self.environ['COMPRESS'])
self.ctx = zmq.Context()
from dpark.cache import CacheTracker, LocalCacheTracker
if isLocal:
self.cacheTracker = LocalCacheTracker(isMaster)
else:
self.cacheTracker = CacheTracker(isMaster)
from dpark.shuffle import LocalFileShuffle, MapOutputTracker, LocalMapOutputTracker
LocalFileShuffle.initialize(isMaster)
if isLocal:
self.mapOutputTracker = LocalMapOutputTracker(isMaster)
else:
self.mapOutputTracker = MapOutputTracker(isMaster)
from dpark.shuffle import SimpleShuffleFetcher, ParallelShuffleFetcher
#self.shuffleFetcher = SimpleShuffleFetcher()
self.shuffleFetcher = ParallelShuffleFetcher(2)
from dpark.broadcast import TheBroadcast
TheBroadcast.initialize(isMaster)
self.started = True
logger.debug("env started")
def stop(self):
if not getattr(self, 'started', False):
return
logger.debug("stop env in %s", os.getpid())
self.shuffleFetcher.stop()
self.cacheTracker.stop()
self.mapOutputTracker.stop()
from dpark.broadcast import TheBroadcast
TheBroadcast.shutdown()
logger.debug("cleaning workdir ...")
for d in self.workdir:
shutil.rmtree(d, True)
logger.debug("done.")
self.started = False
class DparkEnv:
environ = {}
@classmethod
def register(cls, name, value):
cls.environ[name] = value
@classmethod
def get(cls, name, default=None):
return cls.environ.get(name, default)
def __init__(self):
self.started = False
def start(self, isMaster, environ={}):
if self.started:
return
logger.debug("start env in %s: %s %s", os.getpid(), isMaster, environ)
self.isMaster = isMaster
if isMaster:
roots = conf.DPARK_WORK_DIR
if isinstance(roots, str):
roots = roots.split(',')
name = '%s-%s-%d' % (time.strftime("%Y%m%d-%H%M%S"),
socket.gethostname(), os.getpid())
self.workdir = [os.path.join(root, name) for root in roots]
for d in self.workdir:
if not os.path.exists(d):
try:
os.makedirs(d)
except OSError:
pass
self.environ['SERVER_URI'] = 'file://' + self.workdir[0]
self.environ['WORKDIR'] = self.workdir
self.environ['COMPRESS'] = util.COMPRESS
else:
self.environ.update(environ)
if self.environ['COMPRESS'] != util.COMPRESS:
raise Exception("no %s available" % self.environ['COMPRESS'])
self.ctx = zmq.Context()
from dpark.tracker import TrackerServer, TrackerClient
if isMaster:
self.trackerServer = TrackerServer()
self.trackerServer.start()
addr = self.trackerServer.addr
env.register('TrackerAddr', addr)
else:
addr = env.get('TrackerAddr')
self.trackerClient = TrackerClient(addr)
from dpark.cache import CacheTracker
self.cacheTracker = CacheTracker()
from dpark.shuffle import LocalFileShuffle, MapOutputTracker
LocalFileShuffle.initialize(isMaster)
self.mapOutputTracker = MapOutputTracker()
from dpark.shuffle import SimpleShuffleFetcher, ParallelShuffleFetcher
#self.shuffleFetcher = SimpleShuffleFetcher()
self.shuffleFetcher = ParallelShuffleFetcher(2)
from dpark.broadcast import start_manager
start_manager(isMaster)
self.started = True
logger.debug("env started")
def stop(self):
if not getattr(self, 'started', False):
return
logger.debug("stop env in %s", os.getpid())
self.shuffleFetcher.stop()
self.cacheTracker.stop()
self.mapOutputTracker.stop()
if self.isMaster:
self.trackerServer.stop()
from dpark.broadcast import stop_manager
stop_manager()
logger.debug("cleaning workdir ...")
for d in self.workdir:
shutil.rmtree(d, True)
logger.debug("done.")
self.started = False
class DparkEnv:
environ = {}
@classmethod
def register(cls, name, value):
cls.environ[name] = value
@classmethod
def get(cls, name, default=None):
return cls.environ.get(name, default)
def __init__(self):
self.started = False
def start(self, isMaster, environ={}, isLocal=False):
if self.started:
return
logger.debug("start env in %s: %s %s", os.getpid(), isMaster, environ)
self.isMaster = isMaster
self.isLocal = isLocal
if isMaster:
roots = conf.DPARK_WORK_DIR.split(',')
if isLocal:
root = roots[0] # for local mode
if not os.path.exists(root):
os.mkdir(root, 0777)
os.chmod(root, 0777) # because of umask
name = '%s-%s-%d' % (time.strftime("%Y%m%d-%H%M%S"),
socket.gethostname(), os.getpid())
self.workdir = [os.path.join(root, name) for root in roots]
for d in self.workdir:
if not os.path.exists(d):
try:
os.makedirs(d)
except OSError:
pass
self.environ['WORKDIR'] = self.workdir
self.environ['COMPRESS'] = util.COMPRESS
else:
self.environ.update(environ)
if self.environ['COMPRESS'] != util.COMPRESS:
raise Exception("no %s available" % self.environ['COMPRESS'])
self.ctx = zmq.Context()
from dpark.cache import CacheTracker, LocalCacheTracker
if isLocal:
self.cacheTracker = LocalCacheTracker(isMaster)
else:
self.cacheTracker = CacheTracker(isMaster)
from dpark.shuffle import LocalFileShuffle, MapOutputTracker, LocalMapOutputTracker
LocalFileShuffle.initialize(isMaster)
if isLocal:
self.mapOutputTracker = LocalMapOutputTracker(isMaster)
else:
self.mapOutputTracker = MapOutputTracker(isMaster)
from dpark.shuffle import SimpleShuffleFetcher, ParallelShuffleFetcher
#self.shuffleFetcher = SimpleShuffleFetcher()
self.shuffleFetcher = ParallelShuffleFetcher(2)
from dpark.broadcast import TheBroadcast
TheBroadcast.initialize(isMaster)
self.started = True
logger.debug("env started")
def stop(self):
if not getattr(self, 'started', False):
return
logger.debug("stop env in %s", os.getpid())
self.shuffleFetcher.stop()
self.cacheTracker.stop()
self.mapOutputTracker.stop()
from dpark.broadcast import TheBroadcast
TheBroadcast.shutdown()
logger.debug("cleaning workdir ...")
for d in self.workdir:
shutil.rmtree(d, True)
logger.debug("done.")
self.started = False
class DparkEnv:
environ = {}
@classmethod
def register(cls, name, value):
cls.environ[name] = value
@classmethod
def get(cls, name, default=None):
return cls.environ.get(name, default)
def __init__(self):
self.started = False
def start(self, isMaster, environ={}):
if self.started:
return
logger.debug("start env in %s: %s %s", os.getpid(), isMaster, environ)
self.isMaster = isMaster
if isMaster:
roots = conf.DPARK_WORK_DIR
if isinstance(roots, str):
roots = roots.split(',')
name = '%s-%s-%d' % (time.strftime("%Y%m%d-%H%M%S"),
socket.gethostname(), os.getpid())
self.workdir = [os.path.join(root, name) for root in roots]
try:
for d in self.workdir:
util.mkdir_p(d)
except OSError as e:
if environ.get('is_local', False):
raise e
self.environ['SERVER_URI'] = 'file://' + self.workdir[0]
self.environ['WORKDIR'] = self.workdir
self.environ['COMPRESS'] = util.COMPRESS
else:
self.environ.update(environ)
if self.environ['COMPRESS'] != util.COMPRESS:
raise Exception("no %s available" % self.environ['COMPRESS'])
self.ctx = zmq.Context()
from dpark.tracker import TrackerServer, TrackerClient
if isMaster:
self.trackerServer = TrackerServer()
self.trackerServer.start()
addr = self.trackerServer.addr
env.register('TrackerAddr', addr)
else:
addr = env.get('TrackerAddr')
self.trackerClient = TrackerClient(addr)
from dpark.cache import CacheTracker
self.cacheTracker = CacheTracker()
from dpark.shuffle import LocalFileShuffle, MapOutputTracker
LocalFileShuffle.initialize(isMaster)
self.mapOutputTracker = MapOutputTracker()
from dpark.shuffle import ParallelShuffleFetcher
self.shuffleFetcher = ParallelShuffleFetcher(2)
from dpark.broadcast import start_manager
start_manager(isMaster)
self.started = True
logger.debug("env started")
def stop(self):
if not getattr(self, 'started', False):
return
logger.debug("stop env in %s", os.getpid())
self.shuffleFetcher.stop()
self.cacheTracker.stop()
self.mapOutputTracker.stop()
if self.isMaster:
self.trackerServer.stop()
from dpark.broadcast import stop_manager
stop_manager()
logger.debug("cleaning workdir ...")
for d in self.workdir:
shutil.rmtree(d, True)
logger.debug("done.")
self.started = False
class DparkEnv:
environ = {}
trackerServer = None
@classmethod
def register(cls, name, value):
cls.environ[name] = value
@classmethod
def get(cls, name, default=None):
return cls.environ.get(name, default)
def __init__(self):
self.started = False
self.task_stats = TaskStats()
name = self.get('DPARK_ID')
if name is None:
name = '%s-%s' % (socket.gethostname(), uuid.uuid4())
self.register('DPARK_ID', name)
self.workdir = self.get('WORKDIR')
if self.workdir is None:
roots = conf.DPARK_WORK_DIR
if isinstance(roots, str):
roots = roots.split(',')
if not roots:
logger.warning('Cannot get WORKDIR, use temp dir instead.')
roots = [tempfile.gettempdir()]
self.workdir = [os.path.join(root, name) for root in roots]
self.register('WORKDIR', self.workdir)
if 'SERVER_URI' not in self.environ:
self.register('SERVER_URI', 'file://' + self.workdir[0])
compress = self.get('COMPRESS')
if compress is None:
self.register('COMPRESS', util.COMPRESS)
compress = self.get('COMPRESS')
if compress != util.COMPRESS:
raise Exception("no %s available" % compress)
def start(self):
if self.started:
return
self.started = True
logger.debug("start env in %s", os.getpid())
for d in self.workdir:
util.mkdir_p(d)
if 'TRACKER_ADDR' not in self.environ:
from dpark.tracker import TrackerServer
trackerServer = self.trackerServer = TrackerServer()
self.trackerServer.start()
self.register('TRACKER_ADDR', trackerServer.addr)
from dpark.tracker import TrackerClient
addr = self.get('TRACKER_ADDR')
self.trackerClient = TrackerClient(addr)
from dpark.cache import CacheTracker
self.cacheTracker = CacheTracker()
from dpark.shuffle import MapOutputTracker
self.mapOutputTracker = MapOutputTracker()
from dpark.shuffle import ParallelShuffleFetcher
self.shuffleFetcher = ParallelShuffleFetcher(2)
from dpark.broadcast import start_guide_manager, GUIDE_ADDR
if GUIDE_ADDR not in self.environ:
start_guide_manager()
logger.debug("env started")
def stop(self):
if not getattr(self, 'started', False):
return
self.started = False
logger.debug("stop env in %s", os.getpid())
self.trackerClient.stop()
self.shuffleFetcher.stop()
self.cacheTracker.stop()
self.mapOutputTracker.stop()
if self.trackerServer is not None:
self.trackerServer.stop()
self.environ.pop('TRACKER_ADDR', None)
from dpark.broadcast import stop_manager
stop_manager()
logger.debug("cleaning workdir ...")
for d in self.workdir:
shutil.rmtree(d, True)
logger.debug("done.")