def resources(self):
cluster, is_distributed, = self.spec.cluster_def
worker_resources = MXNetSpecification.get_worker_resources(
environment=self.spec.environment,
cluster=cluster,
is_distributed=is_distributed)
ps_resources = MXNetSpecification.get_ps_resources(
environment=self.spec.environment,
cluster=cluster,
is_distributed=is_distributed)
return {
TaskType.MASTER: {
0: self.spec.master_resources
},
TaskType.WORKER: worker_resources,
TaskType.SERVER: ps_resources,
}
def resources(self):
cluster, is_distributed, = self.spec.cluster_def
worker_resources = MXNetSpecification.get_worker_resources(
environment=self.spec.environment,
cluster=cluster,
is_distributed=is_distributed
)
ps_resources = MXNetSpecification.get_ps_resources(
environment=self.spec.environment,
cluster=cluster,
is_distributed=is_distributed
)
return {
TaskType.MASTER: {0: self.spec.master_resources},
TaskType.WORKER: worker_resources,
TaskType.SERVER: ps_resources,
}
def handle_mxnet_experiment(experiment, spawner, response):
# Get the number of jobs this experiment started
master = response[TaskType.MASTER]
job_uuid = master['pod']['metadata']['labels']['job_uuid']
job_uuid = uuid.UUID(job_uuid)
create_job(job_uuid=job_uuid,
experiment=experiment,
definition=get_job_definition(master),
resources=spawner.spec.master_resources)
cluster, is_distributed, = spawner.spec.cluster_def
worker_resources = MXNetSpecification.get_worker_resources(
environment=spawner.spec.environment,
cluster=cluster,
is_distributed=is_distributed
)
server_resources = MXNetSpecification.get_ps_resources(
environment=spawner.spec.environment,
cluster=cluster,
is_distributed=is_distributed
)
for i, worker in enumerate(response[TaskType.WORKER]):
job_uuid = worker['pod']['metadata']['labels']['job_uuid']
job_uuid = uuid.UUID(job_uuid)
create_job(job_uuid=job_uuid,
experiment=experiment,
definition=get_job_definition(worker),
role=TaskType.WORKER,
resources=worker_resources.get(i))
for i, server in enumerate(response[TaskType.SERVER]):
job_uuid = server['pod']['metadata']['labels']['job_uuid']
job_uuid = uuid.UUID(job_uuid)
create_job(job_uuid=job_uuid,
experiment=experiment,
definition=get_job_definition(server),
role=TaskType.SERVER,
resources=server_resources.get(i))
def handle_mxnet_experiment(experiment, spawner, response):
# Get the number of jobs this experiment started
master = response[TaskType.MASTER]
job_uuid = master['pod']['metadata']['labels']['job_uuid']
job_uuid = uuid.UUID(job_uuid)
create_job(job_uuid=job_uuid,
experiment=experiment,
definition=get_job_definition(master),
resources=spawner.spec.master_resources)
cluster, is_distributed, = spawner.spec.cluster_def
worker_resources = MXNetSpecification.get_worker_resources(
environment=spawner.spec.environment,
cluster=cluster,
is_distributed=is_distributed
)
server_resources = MXNetSpecification.get_ps_resources(
environment=spawner.spec.environment,
cluster=cluster,
is_distributed=is_distributed
)
for i, worker in enumerate(response[TaskType.WORKER]):
job_uuid = worker['pod']['metadata']['labels']['job_uuid']
job_uuid = uuid.UUID(job_uuid)
create_job(job_uuid=job_uuid,
experiment=experiment,
definition=get_job_definition(worker),
role=TaskType.WORKER,
resources=worker_resources.get(i))
for i, server in enumerate(response[TaskType.SERVER]):
job_uuid = server['pod']['metadata']['labels']['job_uuid']
job_uuid = uuid.UUID(job_uuid)
create_job(job_uuid=job_uuid,
experiment=experiment,
definition=get_job_definition(server),
role=TaskType.SERVER,
resources=server_resources.get(i))
def test_distributed_mxnet_passes(self):
plxfile = PolyaxonFile(
os.path.abspath('tests/fixtures/distributed_mxnet_file.yml'))
spec = plxfile.specification
assert spec.version == 1
assert spec.project.name == 'project1'
assert isinstance(spec.settings, SettingsConfig)
assert isinstance(spec.settings.logging, LoggingConfig)
assert spec.settings.matrix is None
assert isinstance(spec.environment, EnvironmentConfig)
assert spec.is_runnable
assert spec.framework == Frameworks.MXNET
assert spec.environment.mxnet.n_workers == 5
assert spec.environment.mxnet.n_ps == 10
assert isinstance(spec.environment.resources, PodResourcesConfig)
assert isinstance(spec.environment.resources.cpu, K8SResourcesConfig)
assert spec.environment.resources.cpu.requests == 1
assert spec.environment.resources.cpu.limits == 2
assert isinstance(spec.environment.mxnet.default_worker_resources,
PodResourcesConfig)
assert isinstance(spec.environment.mxnet.default_worker_resources.cpu,
K8SResourcesConfig)
assert spec.environment.mxnet.default_worker_resources.cpu.requests == 3
assert spec.environment.mxnet.default_worker_resources.cpu.limits == 3
assert isinstance(
spec.environment.mxnet.default_worker_resources.memory,
K8SResourcesConfig)
assert spec.environment.mxnet.default_worker_resources.memory.requests == 256
assert spec.environment.mxnet.default_worker_resources.memory.limits == 256
assert isinstance(spec.environment.mxnet.worker_resources[0],
PodResourcesConfig)
assert isinstance(spec.environment.mxnet.worker_resources[0].memory,
K8SResourcesConfig)
assert spec.environment.mxnet.worker_resources[0].index == 3
assert spec.environment.mxnet.worker_resources[
0].memory.requests == 300
assert spec.environment.mxnet.worker_resources[0].memory.limits == 300
assert isinstance(spec.environment.mxnet.default_ps_resources,
PodResourcesConfig)
assert isinstance(spec.environment.mxnet.default_ps_resources.cpu,
K8SResourcesConfig)
assert spec.environment.mxnet.default_ps_resources.cpu.requests == 2
assert spec.environment.mxnet.default_ps_resources.cpu.limits == 4
assert isinstance(spec.environment.mxnet.ps_resources[0],
PodResourcesConfig)
assert isinstance(spec.environment.mxnet.ps_resources[0].memory,
K8SResourcesConfig)
assert spec.environment.mxnet.ps_resources[0].index == 9
assert spec.environment.mxnet.ps_resources[0].memory.requests == 512
assert spec.environment.mxnet.ps_resources[0].memory.limits == 1024
# check that properties for return list of configs and resources is working
cluster, is_distributed = spec.cluster_def
worker_resources = MXNetSpecification.get_worker_resources(
environment=spec.environment,
cluster=cluster,
is_distributed=is_distributed)
assert len(worker_resources) == spec.environment.mxnet.n_workers
assert set(worker_resources.values()) == {
spec.environment.mxnet.default_worker_resources,
spec.environment.mxnet.worker_resources[0]
}
ps_resources = MXNetSpecification.get_ps_resources(
environment=spec.environment,
cluster=cluster,
is_distributed=is_distributed)
assert len(ps_resources) == spec.environment.mxnet.n_ps
assert set(ps_resources.values()) == {
spec.environment.mxnet.default_ps_resources,
spec.environment.mxnet.ps_resources[0]
}
# Check total resources
assert spec.total_resources == {
'cpu': {
'requests': 1 + 3 * 4 + 2 * 9,
'limits': 2 + 3 * 4 + 4 * 9
},
'memory': {
'requests': 300 + 256 * 4 + 512,
'limits': 300 + 256 * 4 + 1024
},
'gpu': None
}
assert spec.cluster_def == ({
TaskType.MASTER: 1,
TaskType.WORKER: 5,
TaskType.SERVER: 10
}, True)