def test_get_local_nodelist_distrib_mode():
mynode = socket.gethostname()
# nodelist_in = ['knl-0020', 'knl-0021', 'knl-0022', 'knl-0036', 'knl-0137', 'knl-0138', 'knl-0139', 'knl-1234']
nodelist_in = ['knl-0020', 'knl-0021', 'knl-0022', 'knl-0036', 'knl-0137', 'knl-0138', 'knl-0139']
with open('node_list', 'w') as f:
for i, node in enumerate(nodelist_in):
f.write(node + '\n')
if i == 3:
f.write(mynode + '\n')
resources = Resources(central_mode=False)
# Spoof current process as each worker and check nodelist.
num_workers = 8
# Test workerID not in local_nodelist [update: This should now work - check removed]
# workerID = 4
# try:
# local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
# except:
# assert 1
# else:
# assert 0
workerID = 5
exp_node = mynode # sname(mynode)
exp_out = [exp_node]
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
assert local_nodelist == exp_out, "local_nodelist returned does not match expected"
num_workers = 1
workerID = 1
exp_out = ['knl-0020', 'knl-0021', 'knl-0022', 'knl-0036', exp_node, 'knl-0137', 'knl-0138', 'knl-0139']
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
assert local_nodelist == exp_out, "local_nodelist returned does not match expected"
num_workers = 4
workerID = 3
exp_out = [exp_node, 'knl-0137']
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
assert local_nodelist == exp_out, "local_nodelist returned does not match expected"
# Sub-node workers
num_workers = 16
workerID = 9
exp_out = [exp_node]
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
assert local_nodelist == exp_out, "local_nodelist returned does not match expected"
workerID = 10
exp_out = [exp_node]
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
assert local_nodelist == exp_out, "local_nodelist returned does not match expected"
os.remove('node_list')
def test_get_local_nodelist_central_mode():
os.environ[
"LIBE_RESOURCES_TEST_NODE_LIST"] = "knl-[0020-0022,0036,0137-0139,1234]"
resources = Resources(nodelist_env_slurm="LIBE_RESOURCES_TEST_NODE_LIST",
central_mode=True)
# Now mock up some more stuff - so consistent
# Spoof current process as each worker and check nodelist.
num_workers = 8
exp_out = [['knl-0020'], ['knl-0021'], ['knl-0022'], ['knl-0036'],
['knl-0137'], ['knl-0138'], ['knl-0139'], ['knl-1234']]
for wrk in range(num_workers):
workerID = wrk + 1
local_nodelist = WorkerResources.get_local_nodelist(
num_workers, workerID, resources)
assert local_nodelist == exp_out[
wrk], "local_nodelist returned does not match expected"
# Spoof current process as each worker and check nodelist.
num_workers = 4
exp_out = [['knl-0020', 'knl-0021'], ['knl-0022', 'knl-0036'],
['knl-0137', 'knl-0138'], ['knl-0139', 'knl-1234']]
for wrk in range(num_workers):
workerID = wrk + 1
local_nodelist = WorkerResources.get_local_nodelist(
num_workers, workerID, resources)
assert local_nodelist == exp_out[
wrk], "local_nodelist returned does not match expected"
# Spoof current process as each worker and check nodelist.
num_workers = 1
exp_out = [[
'knl-0020', 'knl-0021', 'knl-0022', 'knl-0036', 'knl-0137', 'knl-0138',
'knl-0139', 'knl-1234'
]]
for wrk in range(num_workers):
workerID = wrk + 1
local_nodelist = WorkerResources.get_local_nodelist(
num_workers, workerID, resources)
assert local_nodelist == exp_out[
wrk], "local_nodelist returned does not match expected"
# Test the best_split algorithm
num_workers = 3
exp_out = [['knl-0020', 'knl-0021', 'knl-0022'],
['knl-0036', 'knl-0137', 'knl-0138'], ['knl-0139', 'knl-1234']]
for wrk in range(num_workers):
workerID = wrk + 1
local_nodelist = WorkerResources.get_local_nodelist(
num_workers, workerID, resources)
assert local_nodelist == exp_out[
wrk], "local_nodelist returned does not match expected"
def test_get_local_nodelist_distrib_mode_host_not_in_list():
os.environ["LIBE_RESOURCES_TEST_NODE_LIST"] = "knl-[0020-0022,0036,0137-0139,1234]"
resources = Resources(nodelist_env_slurm="LIBE_RESOURCES_TEST_NODE_LIST", central_mode=False)
# Spoof current process as each worker and check nodelist.
num_workers = 4
exp_out = ['knl-0022', 'knl-0036']
# Test running distributed mode without current host in list.
workerID = 2
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
# Now this should work
assert local_nodelist == exp_out, "local_nodelist returned does not match expected"
def test_get_local_nodelist_central_mode_remove_libE_proc():
mynode = socket.gethostname()
nodelist_in = ['knl-0020', 'knl-0021', 'knl-0022', 'knl-0036', 'knl-0137', 'knl-0138', 'knl-0139', 'knl-1234']
with open('node_list', 'w') as f:
for i, node in enumerate(nodelist_in):
f.write(node + '\n')
if i == 3:
f.write(mynode + '\n')
resources = Resources(central_mode=True)
resources.add_comm_info(libE_nodes=[mynode])
# Now mock up some more stuff - so consistent
# Spoof current process as each worker and check nodelist.
num_workers = 8
exp_out = [['knl-0020'], ['knl-0021'], ['knl-0022'], ['knl-0036'],
['knl-0137'], ['knl-0138'], ['knl-0139'], ['knl-1234']]
for wrk in range(num_workers):
workerID = wrk + 1
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
assert local_nodelist == exp_out[wrk], "local_nodelist returned does not match expected"
# Spoof current process as each worker and check nodelist.
num_workers = 4
exp_out = [['knl-0020', 'knl-0021'], ['knl-0022', 'knl-0036'], ['knl-0137', 'knl-0138'], ['knl-0139', 'knl-1234']]
for wrk in range(num_workers):
workerID = wrk + 1
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
assert local_nodelist == exp_out[wrk], "local_nodelist returned does not match expected"
# Spoof current process as each worker and check nodelist.
num_workers = 1
exp_out = [['knl-0020', 'knl-0021', 'knl-0022', 'knl-0036', 'knl-0137', 'knl-0138', 'knl-0139', 'knl-1234']]
for wrk in range(num_workers):
workerID = wrk + 1
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
assert local_nodelist == exp_out[wrk], "local_nodelist returned does not match expected"
# Test the best_split algorithm
num_workers = 3
exp_out = [['knl-0020', 'knl-0021', 'knl-0022'], ['knl-0036', 'knl-0137', 'knl-0138'], ['knl-0139', 'knl-1234']]
for wrk in range(num_workers):
workerID = wrk + 1
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
assert local_nodelist == exp_out[wrk], "local_nodelist returned does not match expected"
os.remove('node_list')
def test_worker_resources():
os.environ[
"LIBE_RESOURCES_TEST_NODE_LIST"] = "knl-[0020-0022,0036,0137-0139,1234]"
resources = Resources(nodelist_env_slurm="LIBE_RESOURCES_TEST_NODE_LIST",
central_mode=True)
# One worker per node
exp_nodelist1 = [['knl-0020'], ['knl-0021'], ['knl-0022'], ['knl-0036'],
['knl-0137'], ['knl-0138'], ['knl-0139'], ['knl-1234']]
num_workers = 8
comm = Fake_comm(num_workers)
for wrk in range(num_workers):
workerID = wrk + 1
worker = WorkerResources(workerID, comm, resources)
assert worker.num_workers == 8, 'worker.num_workers does not match'
assert worker.workerID == workerID, 'worker.workerID does not match'
assert worker.local_nodelist == exp_nodelist1[
wrk], 'worker.local_nodelist does not match'
assert worker.local_node_count == 1, 'worker.local_node_count does not match'
assert worker.workers_per_node == 1, 'worker.workers_per_node does not match'
# Multiple nodes per worker
exp_nodelist2 = [['knl-0020', 'knl-0021', 'knl-0022', 'knl-0036'],
['knl-0137', 'knl-0138', 'knl-0139', 'knl-1234']]
num_workers = 2
comm2 = Fake_comm(num_workers)
for wrk in range(num_workers):
workerID = wrk + 1
worker = WorkerResources(workerID, comm2, resources)
assert worker.num_workers == 2, 'worker.num_workers does not match'
assert worker.workerID == workerID, 'worker.workerID does not match'
assert worker.local_nodelist == exp_nodelist2[
wrk], 'worker.local_nodelist does not match'
assert worker.local_node_count == 4, 'worker.local_node_count does not match'
assert worker.workers_per_node == 1, 'worker.workers_per_node does not match'
# Multiple workers per node
num_workers = 16
comm3 = Fake_comm(num_workers)
for wrk in range(num_workers):
workerID = wrk + 1
worker = WorkerResources(workerID, comm3, resources)
assert worker.num_workers == 16, 'worker.num_workers does not match'
assert worker.workerID == workerID, 'worker.workerID does not match'
assert worker.local_nodelist == exp_nodelist1[
wrk // 2], 'worker.local_nodelist does not match'
assert worker.local_node_count == 1, 'worker.local_node_count does not match'
assert worker.workers_per_node == 2, 'worker.workers_per_node does not match'
def test_get_local_nodelist_distrib_mode_uneven_split():
mynode = socket.gethostname()
exp_node = mynode # sname(mynode)
nodelist_in = ['knl-0020', 'knl-0021', 'knl-0022', 'knl-0036', 'knl-0137', 'knl-0138', 'knl-0139', 'knl-1234']
with open('node_list', 'w') as f:
for i, node in enumerate(nodelist_in):
f.write(node + '\n')
if i == 4:
f.write(mynode + '\n')
resources = Resources(central_mode=False)
num_workers = 2
# May not be at head of list - should perhaps be warning or enforced
workerID = 2
exp_out = ['knl-0137', exp_node, 'knl-0138', 'knl-0139']
local_nodelist = WorkerResources.get_local_nodelist(num_workers, workerID, resources)
assert local_nodelist == exp_out, "local_nodelist returned does not match expected"
os.remove('node_list')
def set_worker_info(self, comm, workerid=None):
"""Sets info for this executor"""
self.workerID = workerid
if not self.resources:
self.resources = Resources()
self.resources.set_worker_resources(self.workerID, comm)