def test_service_hosts():
pytest.importorskip("bokeh")
from distributed.bokeh.scheduler import BokehScheduler
port = 0
for url, expected in [
("tcp://0.0.0.0", ("::", "0.0.0.0")),
("tcp://127.0.0.1", "127.0.0.1"),
("tcp://127.0.0.1:38275", "127.0.0.1"),
]:
services = {("bokeh", port): BokehScheduler}
s = Scheduler(services=services)
yield s.start(url)
sock = first(s.services["bokeh"].server._http._sockets.values())
if isinstance(expected, tuple):
assert sock.getsockname()[0] in expected
else:
assert sock.getsockname()[0] == expected
yield s.close()
port = ("127.0.0.1", 0)
for url in ["tcp://0.0.0.0", "tcp://127.0.0.1", "tcp://127.0.0.1:38275"]:
services = {("bokeh", port): BokehScheduler}
s = Scheduler(services=services)
yield s.start(url)
sock = first(s.services["bokeh"].server._http._sockets.values())
assert sock.getsockname()[0] == "127.0.0.1"
yield s.close()
def test_persist_taskstate():
s = Scheduler(validate=True, persist_file='persist_test')
s.start(0)
assert s.persist_scheduler
s.update_graph(tasks={
'x': dumps_task((inc, 1)),
'y': dumps_task((inc, 'x')),
'z': dumps_task((inc, 2))
},
keys=['y'],
dependencies={
'y': 'x',
'x': [],
'z': []
},
client='client')
taskstates = s.tasks
s.close()
s.stop()
del s
s = Scheduler(validate=True, persist_file='persist_test')
s.start(0)
assert ([taskstates.keys()] == [s.tasks.keys()]
and [x.state for x in taskstates.values()
] == [x.state for x in s.tasks.values()])
s.close()
for f in glob.glob("persist_test*"):
os.remove(f)
def test_service_hosts():
pytest.importorskip('bokeh')
from distributed.bokeh.scheduler import BokehScheduler
port = 0
for url, expected in [('tcp://0.0.0.0', ('::', '0.0.0.0')),
('tcp://127.0.0.1', '127.0.0.1'),
('tcp://127.0.0.1:38275', '127.0.0.1')]:
services = {('bokeh', port): BokehScheduler}
s = Scheduler(services=services)
yield s.start(url)
sock = first(s.services['bokeh'].server._http._sockets.values())
if isinstance(expected, tuple):
assert sock.getsockname()[0] in expected
else:
assert sock.getsockname()[0] == expected
yield s.close()
port = ('127.0.0.1', 0)
for url in ['tcp://0.0.0.0', 'tcp://127.0.0.1', 'tcp://127.0.0.1:38275']:
services = {('bokeh', port): BokehScheduler}
s = Scheduler(services=services)
yield s.start(url)
sock = first(s.services['bokeh'].server._http._sockets.values())
assert sock.getsockname()[0] == '127.0.0.1'
yield s.close()
def test_launch_without_blocked_services():
from distributed.http import HTTPScheduler
s = Scheduler(services={('http', 3849): HTTPScheduler})
s.start(0)
s2 = Scheduler(services={('http', 3849): HTTPScheduler})
s2.start(0)
assert not s2.services
yield [s.close(), s2.close()]
def test_launch_without_blocked_services():
from distributed.http import HTTPScheduler
s = Scheduler(services={('http', 3849): HTTPScheduler})
s.start(0)
s2 = Scheduler(services={('http', 3849): HTTPScheduler})
s2.start(0)
assert not s2.services
yield [s.close(), s2.close()]
def test_scheduler_as_center():
s = Scheduler()
s.listen(0)
done = s.start()
a = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=1)
a.data.update({'x': 1, 'y': 2})
b = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=2)
b.data.update({'y': 2, 'z': 3})
c = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=3)
yield [w._start() for w in [a, b, c]]
assert s.ncores == {w.address: w.ncores for w in [a, b, c]}
assert s.who_has == {
'x': {a.address},
'y': {a.address, b.address},
'z': {b.address}
}
s.update_graph(dsk={'a': (inc, 1)}, keys=['a'])
while not s.who_has['a']:
yield gen.sleep(0.01)
assert 'a' in a.data or 'a' in b.data or 'a' in c.data
yield [w._close() for w in [a, b, c]]
assert s.ncores == {}
assert s.who_has == {}
yield s.close()
def test_scheduler_as_center():
s = Scheduler()
done = s.start(0)
a = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=1)
a.data.update({'x': 1, 'y': 2})
b = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=2)
b.data.update({'y': 2, 'z': 3})
c = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=3)
yield [w._start(0) for w in [a, b, c]]
assert s.ncores == {w.address: w.ncores for w in [a, b, c]}
assert s.who_has == {
'x': {a.address},
'y': {a.address, b.address},
'z': {b.address}
}
s.update_graph(tasks={'a': dumps_task((inc, 1))},
keys=['a'],
dependencies={'a': []})
start = time()
while not s.who_has['a']:
assert time() - start < 5
yield gen.sleep(0.01)
assert 'a' in a.data or 'a' in b.data or 'a' in c.data
with ignoring(StreamClosedError):
yield [w._close() for w in [a, b, c]]
assert s.ncores == {}
assert s.who_has == {}
yield s.close()
def test_scheduler_as_center():
s = Scheduler()
s.listen(0)
done = s.start()
a = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=1)
a.data.update({'x': 1, 'y': 2})
b = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=2)
b.data.update({'y': 2, 'z': 3})
c = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=3)
yield [w._start() for w in [a, b, c]]
assert s.ncores == {w.address: w.ncores for w in [a, b, c]}
assert s.who_has == {'x': {a.address},
'y': {a.address, b.address},
'z': {b.address}}
s.update_graph(dsk={'a': (inc, 1)},
keys=['a'])
while not s.who_has['a']:
yield gen.sleep(0.01)
assert 'a' in a.data or 'a' in b.data or 'a' in c.data
yield [w._close() for w in [a, b, c]]
assert s.ncores == {}
assert s.who_has == {}
yield s.close()
def test_scheduler_as_center():
s = Scheduler(validate=True)
done = s.start(0)
a = Worker(s.address, ncores=1)
a.data.update({'x': 1, 'y': 2})
b = Worker(s.address, ncores=2)
b.data.update({'y': 2, 'z': 3})
c = Worker(s.address, ncores=3)
yield [w._start(0) for w in [a, b, c]]
assert s.ncores == {w.address: w.ncores for w in [a, b, c]}
assert not s.who_has
s.update_graph(tasks={'a': dumps_task((inc, 1))},
keys=['a'],
dependencies={'a': []})
start = time()
while not 'a' in s.who_has:
assert time() - start < 5
yield gen.sleep(0.01)
assert 'a' in a.data or 'a' in b.data or 'a' in c.data
yield [w._close() for w in [a, b, c]]
assert s.ncores == {}
assert s.who_has == {}
yield s.close()
def test_coerce_address():
with dask.config.set({'distributed.comm.timeouts.connect': '100ms'}):
s = Scheduler(validate=True)
s.start(0)
print("scheduler:", s.address, s.listen_address)
a = Worker(s.ip, s.port, name='alice')
b = Worker(s.ip, s.port, name=123)
c = Worker('127.0.0.1', s.port, name='charlie')
yield [a._start(), b._start(), c._start()]
assert s.coerce_address('127.0.0.1:8000') == 'tcp://127.0.0.1:8000'
assert s.coerce_address('[::1]:8000') == 'tcp://[::1]:8000'
assert s.coerce_address('tcp://127.0.0.1:8000') == 'tcp://127.0.0.1:8000'
assert s.coerce_address('tcp://[::1]:8000') == 'tcp://[::1]:8000'
assert s.coerce_address('localhost:8000') in ('tcp://127.0.0.1:8000', 'tcp://[::1]:8000')
assert s.coerce_address(u'localhost:8000') in ('tcp://127.0.0.1:8000', 'tcp://[::1]:8000')
assert s.coerce_address(a.address) == a.address
# Aliases
assert s.coerce_address('alice') == a.address
assert s.coerce_address(123) == b.address
assert s.coerce_address('charlie') == c.address
assert s.coerce_hostname('127.0.0.1') == '127.0.0.1'
assert s.coerce_hostname('alice') == a.ip
assert s.coerce_hostname(123) == b.ip
assert s.coerce_hostname('charlie') == c.ip
assert s.coerce_hostname('jimmy') == 'jimmy'
assert s.coerce_address('zzzt:8000', resolve=False) == 'tcp://zzzt:8000'
yield s.close()
yield [w._close() for w in [a, b, c]]
def test_coerce_address():
s = Scheduler(validate=True)
s.start(0)
print("scheduler:", s.address, s.listen_address)
a = Worker(s.ip, s.port, name='alice')
b = Worker(s.ip, s.port, name=123)
c = Worker('127.0.0.1', s.port, name='charlie')
yield [a._start(), b._start(), c._start()]
assert s.coerce_address('127.0.0.1:8000') == 'tcp://127.0.0.1:8000'
assert s.coerce_address('[::1]:8000') == 'tcp://[::1]:8000'
assert s.coerce_address('tcp://127.0.0.1:8000') == 'tcp://127.0.0.1:8000'
assert s.coerce_address('tcp://[::1]:8000') == 'tcp://[::1]:8000'
assert s.coerce_address('localhost:8000') in ('tcp://127.0.0.1:8000', 'tcp://[::1]:8000')
assert s.coerce_address(u'localhost:8000') in ('tcp://127.0.0.1:8000', 'tcp://[::1]:8000')
assert s.coerce_address(a.address) == a.address
# Aliases
assert s.coerce_address('alice') == a.address
assert s.coerce_address(123) == b.address
assert s.coerce_address('charlie') == c.address
assert s.coerce_hostname('127.0.0.1') == '127.0.0.1'
assert s.coerce_hostname('alice') == a.ip
assert s.coerce_hostname(123) == b.ip
assert s.coerce_hostname('charlie') == c.ip
assert s.coerce_hostname('jimmy') == 'jimmy'
assert s.coerce_address('zzzt:8000', resolve=False) == 'tcp://zzzt:8000'
yield s.close()
yield [w._close() for w in [a, b, c]]
def test_scheduler_as_center():
s = Scheduler()
done = s.start(0)
a = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=1)
a.data.update({'x': 1, 'y': 2})
b = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=2)
b.data.update({'y': 2, 'z': 3})
c = Worker('127.0.0.1', s.port, ip='127.0.0.1', ncores=3)
yield [w._start(0) for w in [a, b, c]]
assert s.ncores == {w.address: w.ncores for w in [a, b, c]}
assert s.who_has == {'x': {a.address},
'y': {a.address, b.address},
'z': {b.address}}
s.update_graph(tasks={'a': dumps_task((inc, 1))},
keys=['a'],
dependencies={'a': []})
start = time()
while not s.who_has['a']:
assert time() - start < 5
yield gen.sleep(0.01)
assert 'a' in a.data or 'a' in b.data or 'a' in c.data
with ignoring(StreamClosedError):
yield [w._close() for w in [a, b, c]]
assert s.ncores == {}
assert s.who_has == {}
yield s.close()
def test_service_hosts_match_scheduler():
from distributed.http.scheduler import HTTPScheduler
services = {('http', 0): HTTPScheduler}
s = Scheduler(services=services)
yield s.start('tcp://0.0.0.0')
sock = first(s.services['http']._sockets.values())
assert sock.getsockname()[0] in ('::', '0.0.0.0')
yield s.close()
for host in ['tcp://127.0.0.2', 'tcp://127.0.0.2:38275']:
s = Scheduler(services=services)
yield s.start(host)
sock = first(s.services['http']._sockets.values())
assert sock.getsockname()[0] == '127.0.0.2'
yield s.close()
def test_service_hosts_match_scheduler():
pytest.importorskip('bokeh')
from distributed.bokeh.scheduler import BokehScheduler
services = {('bokeh', 0): BokehScheduler}
s = Scheduler(services=services)
yield s.start('tcp://0.0.0.0')
sock = first(s.services['bokeh'].server._http._sockets.values())
assert sock.getsockname()[0] in ('::', '0.0.0.0')
yield s.close()
for host in ['tcp://127.0.0.2', 'tcp://127.0.0.2:38275']:
s = Scheduler(services=services)
yield s.start(host)
sock = first(s.services['bokeh'].server._http._sockets.values())
assert sock.getsockname()[0] == '127.0.0.2'
yield s.close()
def test_scheduler_file():
with tmpfile() as fn:
s = Scheduler(scheduler_file=fn)
s.start(0)
with open(fn) as f:
data = json.load(f)
assert data['address'] == s.address
c = yield Client(scheduler_file=fn, loop=s.loop, asynchronous=True)
yield s.close()
def test_scheduler_file():
with tmpfile() as fn:
s = Scheduler(scheduler_file=fn)
s.start(0)
with open(fn) as f:
data = json.load(f)
assert data['address'] == s.address
c = yield Client(scheduler_file=fn, loop=s.loop, asynchronous=True)
yield s.close()
def test_scheduler_file():
with tmpfile() as fn:
s = Scheduler(scheduler_file=fn)
s.start(0)
with open(fn) as f:
data = json.load(f)
assert data['address'] == s.address
c = Client(scheduler_file=fn, loop=s.loop, start=False)
yield c._start()
yield s.close()
def test_file_descriptors_dont_leak(loop):
psutil = pytest.importorskip('psutil')
proc = psutil.Process()
before = proc.num_fds()
s = Scheduler()
s.start(0)
w = Worker(s.ip, s.port)
@gen.coroutine
def f():
yield w._start(0)
yield w._close()
loop.run_sync(f)
during = proc.num_fds()
s.stop()
s.close()
start = time()
while proc.num_fds() > before:
loop.run_sync(lambda: gen.sleep(0.01))
assert time() < start + 5
def test_worker_name():
s = Scheduler(validate=True)
s.start(0)
w = yield Worker(s.ip, s.port, name="alice")
assert s.workers[w.address].name == "alice"
assert s.aliases["alice"] == w.address
with pytest.raises(ValueError):
w2 = yield Worker(s.ip, s.port, name="alice")
yield w2._close()
yield s.close()
yield w._close()
def test_file_descriptors_dont_leak(loop):
psutil = pytest.importorskip('psutil')
proc = psutil.Process()
before = proc.num_fds()
s = Scheduler()
s.start(0)
w = Worker(s.ip, s.port)
@gen.coroutine
def f():
yield w._start(0)
yield w._close()
loop.run_sync(f)
during = proc.num_fds()
s.stop()
s.close()
start = time()
while proc.num_fds() > before:
loop.run_sync(lambda: gen.sleep(0.01))
assert time() < start + 5
def test_worker_name():
s = Scheduler(validate=True)
s.start(0)
w = Worker(s.ip, s.port, name='alice')
yield w._start()
assert s.worker_info[w.address]['name'] == 'alice'
assert s.aliases['alice'] == w.address
with pytest.raises(ValueError):
w = Worker(s.ip, s.port, name='alice')
yield w._start()
yield s.close()
yield w._close()
def test_worker_name():
s = Scheduler()
s.start(0)
w = Worker(s.ip, s.port, name='alice')
yield w._start()
assert s.worker_info[w.address]['name'] == 'alice'
assert s.aliases['alice'] == w.address
with pytest.raises(ValueError):
w = Worker(s.ip, s.port, name='alice')
yield w._start()
yield s.close()
yield w._close()
def test_worker_name():
s = Scheduler(validate=True)
s.start(0)
w = Worker(s.ip, s.port, name='alice')
yield w._start()
assert s.workers[w.address].name == 'alice'
assert s.aliases['alice'] == w.address
with pytest.raises(ValueError):
w2 = Worker(s.ip, s.port, name='alice')
yield w2._start()
yield w2._close()
yield s.close()
yield w._close()
def test_service_hosts():
pytest.importorskip('bokeh')
from distributed.bokeh.scheduler import BokehScheduler
for port in [0, ('127.0.0.3', 0)]:
for url, expected in [('tcp://0.0.0.0', ('::', '0.0.0.0')),
('tcp://127.0.0.2', '127.0.0.2'),
('tcp://127.0.0.2:38275', '127.0.0.2')]:
services = {('bokeh', port): BokehScheduler}
s = Scheduler(services=services)
yield s.start(url)
sock = first(s.services['bokeh'].server._http._sockets.values())
if isinstance(port, tuple): # host explicitly overridden
assert sock.getsockname()[0] == port[0]
elif isinstance(expected, tuple):
assert sock.getsockname()[0] in expected
else:
assert sock.getsockname()[0] == expected
yield s.close()
def test_service_hosts():
pytest.importorskip('bokeh')
from distributed.bokeh.scheduler import BokehScheduler
for port in [0, ('127.0.0.3', 0)]:
for url, expected in [('tcp://0.0.0.0', ('::', '0.0.0.0')),
('tcp://127.0.0.2', '127.0.0.2'),
('tcp://127.0.0.2:38275', '127.0.0.2')]:
services = {('bokeh', port): BokehScheduler}
s = Scheduler(services=services)
yield s.start(url)
sock = first(s.services['bokeh'].server._http._sockets.values())
if isinstance(port, tuple): # host explicitly overridden
assert sock.getsockname()[0] == port[0]
elif isinstance(expected, tuple):
assert sock.getsockname()[0] in expected
else:
assert sock.getsockname()[0] == expected
yield s.close()
def test_coerce_address():
with dask.config.set({"distributed.comm.timeouts.connect": "100ms"}):
s = Scheduler(validate=True)
s.start(0)
print("scheduler:", s.address, s.listen_address)
a = Worker(s.ip, s.port, name="alice")
b = Worker(s.ip, s.port, name=123)
c = Worker("127.0.0.1", s.port, name="charlie")
yield [a, b, c]
assert s.coerce_address("127.0.0.1:8000") == "tcp://127.0.0.1:8000"
assert s.coerce_address("[::1]:8000") == "tcp://[::1]:8000"
assert s.coerce_address(
"tcp://127.0.0.1:8000") == "tcp://127.0.0.1:8000"
assert s.coerce_address("tcp://[::1]:8000") == "tcp://[::1]:8000"
assert s.coerce_address("localhost:8000") in (
"tcp://127.0.0.1:8000",
"tcp://[::1]:8000",
)
assert s.coerce_address(u"localhost:8000") in (
"tcp://127.0.0.1:8000",
"tcp://[::1]:8000",
)
assert s.coerce_address(a.address) == a.address
# Aliases
assert s.coerce_address("alice") == a.address
assert s.coerce_address(123) == b.address
assert s.coerce_address("charlie") == c.address
assert s.coerce_hostname("127.0.0.1") == "127.0.0.1"
assert s.coerce_hostname("alice") == a.ip
assert s.coerce_hostname(123) == b.ip
assert s.coerce_hostname("charlie") == c.ip
assert s.coerce_hostname("jimmy") == "jimmy"
assert s.coerce_address("zzzt:8000",
resolve=False) == "tcp://zzzt:8000"
yield s.close()
yield [w._close() for w in [a, b, c]]
def test_coerce_address():
s = Scheduler()
s.start(0)
a = Worker(s.ip, s.port, name='alice')
b = Worker(s.ip, s.port, name=123)
c = Worker(s.ip, s.port, name='charlie', ip='127.0.0.2')
yield [a._start(), b._start(), c._start()]
assert s.coerce_address(b'127.0.0.1') == '127.0.0.1'
assert s.coerce_address(('127.0.0.1', 8000)) == '127.0.0.1:8000'
assert s.coerce_address(['127.0.0.1', 8000]) == '127.0.0.1:8000'
assert s.coerce_address([b'127.0.0.1', 8000]) == '127.0.0.1:8000'
assert s.coerce_address(('127.0.0.1', '8000')) == '127.0.0.1:8000'
assert s.coerce_address(b'localhost') == '127.0.0.1'
assert s.coerce_address('localhost') == '127.0.0.1'
assert s.coerce_address(u'localhost') == '127.0.0.1'
assert s.coerce_address('localhost:8000') == '127.0.0.1:8000'
assert s.coerce_address(a.address) == a.address
assert s.coerce_address(a.address_tuple) == a.address
assert s.coerce_address(123) == b.address
assert s.coerce_address('charlie') == c.address
yield s.close()
yield [w._close() for w in [a, b, c]]
def test_coerce_address():
s = Scheduler(validate=True)
s.start(0)
a = Worker(s.ip, s.port, name='alice')
b = Worker(s.ip, s.port, name=123)
c = Worker(s.ip, s.port, name='charlie', ip='127.0.0.2')
yield [a._start(), b._start(), c._start()]
assert s.coerce_address(b'127.0.0.1') == '127.0.0.1'
assert s.coerce_address(('127.0.0.1', 8000)) == '127.0.0.1:8000'
assert s.coerce_address(['127.0.0.1', 8000]) == '127.0.0.1:8000'
assert s.coerce_address([b'127.0.0.1', 8000]) == '127.0.0.1:8000'
assert s.coerce_address(('127.0.0.1', '8000')) == '127.0.0.1:8000'
assert s.coerce_address(b'localhost') == '127.0.0.1'
assert s.coerce_address('localhost') == '127.0.0.1'
assert s.coerce_address(u'localhost') == '127.0.0.1'
assert s.coerce_address('localhost:8000') == '127.0.0.1:8000'
assert s.coerce_address(a.address) == a.address
assert s.coerce_address(a.address_tuple) == a.address
assert s.coerce_address(123) == b.address
assert s.coerce_address('charlie') == c.address
yield s.close()
yield [w._close() for w in [a, b, c]]