def create_pool(self, size, name):
"""Create and deploy a new pool.
Called on job submission by submission.py.
TODO: Support auto-scale formula.
"""
image = self.environment.get_image()
node_agent_sku_id = image.pop('node_sku_id')
pool_id = 'Maya_Pool_{}'.format(uuid.uuid4())
pool_config = models.VirtualMachineConfiguration(
image_reference=models.ImageReference(**image),
node_agent_sku_id=node_agent_sku_id)
self._log.info("Creating new pool '{}' with {} VMs.".format(
name, size))
new_pool = models.PoolAddParameter(
id=pool_id,
display_name="Maya Pool for {}".format(name),
resize_timeout=datetime.timedelta(minutes=30),
application_licenses=self.environment.get_application_licenses(),
vm_size=self.environment.get_vm_sku(),
virtual_machine_configuration=pool_config,
target_dedicated_nodes=int(size[0]),
target_low_priority_nodes=int(size[1]),
max_tasks_per_node=1)
self._call(self.batch.pool.add, new_pool)
self._log.debug("Successfully created pool.")
return {"poolId": pool_id}
def create_pool(batch_client,
name_pool,
number_nodes=0,
cmd_s_task=None,
rule_scale_pool=None):
#parameter image node
param_image = models.VirtualMachineConfiguration(
image_reference=models.ImageReference(offer='UbuntuServer',
publisher='Canonical',
sku='18.04-LTS',
version='latest',
virtual_machine_image_id=None),
node_agent_sku_id='batch.node.ubuntu 18.04')
#parameter pool
new_pool = models.PoolAddParameter(
id=name_pool,
vm_size='standard_d1_v2',
#target_dedicated_nodes = number_nodes,
virtual_machine_configuration=param_image,
enable_inter_node_communication=True,
enable_auto_scale=True,
auto_scale_formula=rule_scale_pool,
auto_scale_evaluation_interval='PT5M'
#start_task = s_task
)
batch_client.pool.add(new_pool)
def test_batch_create_pool_with_blobfuse_mount(self, **kwargs):
client = self.create_sharedkey_client(**kwargs)
# Test Create Iaas Pool
test_iaas_pool = models.PoolAddParameter(
id=self.get_resource_name('batch_iaas_'),
vm_size='Standard_A1',
virtual_machine_configuration=models.VirtualMachineConfiguration(
image_reference=models.ImageReference(
publisher='MicrosoftWindowsServer',
offer='WindowsServer',
sku='2016-Datacenter-smalldisk'
),
node_agent_sku_id='batch.node.windows amd64',
windows_configuration=models.WindowsConfiguration(enable_automatic_updates=True)),
task_scheduling_policy=models.TaskSchedulingPolicy(node_fill_type=models.ComputeNodeFillType.pack),
mount_configuration=[models.MountConfiguration(
azure_blob_file_system_configuration=models.AzureBlobFileSystemConfiguration(
account_name='test',
container_name='https://test.blob.core.windows.net:443/test-container',
relative_mount_path='foo',
account_key='fake_account_key'
)
)]
)
response = client.pool.add(test_iaas_pool)
self.assertIsNone(response)
mount_pool = client.pool.get(test_iaas_pool.id)
self.assertIsNotNone(mount_pool.mount_configuration)
self.assertEqual(len(mount_pool.mount_configuration), 1)
self.assertIsNotNone(mount_pool.mount_configuration[0].azure_blob_file_system_configuration)
self.assertIsNone(mount_pool.mount_configuration[0].nfs_mount_configuration)
def create_pool(batch_client, block_blob_client, pool_id, vm_size, vm_count):
"""Creates an Azure Batch pool with the specified id.
:param batch_client: The batch client to use.
:type batch_client: `batchserviceclient.BatchServiceClient`
:param block_blob_client: The storage block blob client to use.
:type block_blob_client: `azure.storage.blob.BlockBlobService`
:param str pool_id: The id of the pool to create.
:param str vm_size: vm size (sku)
:param int vm_count: number of vms to allocate
"""
#application_packages = [
# batchmodels.ApplicationPackageReference(application_id='Regridv3', version='1.0.0'),
# batchmodels.ApplicationPackageReference(application_id='us_5km', version='1.0.0'),
# batchmodels.ApplicationPackageReference(application_id='us_800m_t6', version='1.0.0'),
#]
sku_to_use, image_ref_to_use = \
common.helpers.select_latest_verified_vm_image_with_node_agent_sku(
batch_client, 'Canonical', 'UbuntuServer', '18.04')
#image_ref_to_use = batchmodels.ImageReference(
# publisher = 'microsoftwindowsserver', offer = 'windowsserver', sku = '2019-datacenter', version = 'latest'
#)
#sku_to_use = "batch.node.windows amd64"
pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use, node_agent_sku_id=sku_to_use),
vm_size=vm_size,
max_tasks_per_node=_POOL_MAX_TASKS_PER_NODE,
#application_package_references=application_packages,
target_dedicated_nodes=vm_count)
common.helpers.create_pool_if_not_exist(batch_client, pool)
def __create_pool_and_job(self, cluster_conf: models.ClusterConfiguration, software_metadata_key: str, start_task, VmImageModel):
"""
Create a pool and job
:param cluster_conf: the configuration object used to create the cluster
:type cluster_conf: aztk.models.ClusterConfiguration
:parm software_metadata_key: the id of the software being used on the cluster
:param start_task: the start task for the cluster
:param VmImageModel: the type of image to provision for the cluster
:param wait: wait until the cluster is ready
"""
self._get_cluster_data(cluster_conf.cluster_id).save_cluster_config(cluster_conf)
# reuse pool_id as job_id
pool_id = cluster_conf.cluster_id
job_id = cluster_conf.cluster_id
# Get a verified node agent sku
sku_to_use, image_ref_to_use = \
helpers.select_latest_verified_vm_image_with_node_agent_sku(
VmImageModel.publisher, VmImageModel.offer, VmImageModel.sku, self.batch_client)
network_conf = None
if cluster_conf.subnet_id is not None:
network_conf = batch_models.NetworkConfiguration(
subnet_id=cluster_conf.subnet_id)
auto_scale_formula = "$TargetDedicatedNodes={0}; $TargetLowPriorityNodes={1}".format(
cluster_conf.vm_count, cluster_conf.vm_low_pri_count)
# Confiure the pool
pool = batch_models.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batch_models.VirtualMachineConfiguration(
image_reference=image_ref_to_use,
node_agent_sku_id=sku_to_use),
vm_size=cluster_conf.vm_size,
enable_auto_scale=True,
auto_scale_formula=auto_scale_formula,
auto_scale_evaluation_interval=timedelta(minutes=5),
start_task=start_task,
enable_inter_node_communication=True if not cluster_conf.subnet_id else False,
max_tasks_per_node=4,
network_configuration=network_conf,
metadata=[
batch_models.MetadataItem(
name=constants.AZTK_SOFTWARE_METADATA_KEY, value=software_metadata_key),
batch_models.MetadataItem(
name=constants.AZTK_MODE_METADATA_KEY, value=constants.AZTK_CLUSTER_MODE_METADATA)
])
# Create the pool + create user for the pool
helpers.create_pool_if_not_exist(pool, self.batch_client)
# Create job
job = batch_models.JobAddParameter(
id=job_id,
pool_info=batch_models.PoolInformation(pool_id=pool_id))
# Add job to batch
self.batch_client.job.add(job)
return helpers.get_cluster(cluster_conf.cluster_id, self.batch_client)
def create_pool(batch_service_client: BatchServiceClient, pool_id: str):
"""
Creates a pool of compute nodes with the specified OS settings.
:param batch_service_client: A Batch service client.
:param str pool_id: An ID for the new pool.
:param str publisher: Marketplace image publisher
:param str offer: Marketplace image offer
:param str sku: Marketplace image sku
"""
print(f'Creating pool [{pool_id}]...')
# Create a new pool of Linux compute nodes using an Azure Virtual Machines
# Marketplace image. For more information about creating pools of Linux
# nodes, see:
# https://azure.microsoft.com/documentation/articles/batch-linux-nodes/
new_pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=batchmodels.ImageReference(
publisher="canonical",
offer="0001-com-ubuntu-server-focal",
sku="20_04-lts",
version="latest"),
node_agent_sku_id="batch.node.ubuntu 20.04"),
vm_size=config.POOL_VM_SIZE,
target_dedicated_nodes=config.POOL_NODE_COUNT)
batch_service_client.pool.add(new_pool)
def create_pool(batch_client, pool_id, vm_size, vm_count, app_files):
"""Creates an Azure Batch pool with the specified id.
:param batch_client: The batch client to use.
:type batch_client: `batchserviceclient.BatchServiceClient`
:param block_blob_client: The storage block blob client to use.
:type block_blob_client: `azure.storage.blob.BlockBlobService`
:param str pool_id: The id of the pool to create.
:param str vm_size: vm size (sku)
:param int vm_count: number of vms to allocate
:param list app_files: The list of all the other scripts to upload.
"""
# pick the latest supported 16.04 sku for UbuntuServer
sku_to_use, image_ref_to_use = \
common.helpers.select_latest_verified_vm_image_with_node_agent_sku(
batch_client, 'Canonical', 'UbuntuServer', '14.04')
user = batchmodels.AutoUserSpecification(
scope=batchmodels.AutoUserScope.pool,
elevation_level=batchmodels.ElevationLevel.admin)
task_commands = get_list_from_file('configs/start_commands')
print(task_commands)
pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use, node_agent_sku_id=sku_to_use),
vm_size=vm_size,
target_dedicated=vm_count,
start_task=batchmodels.StartTask(
command_line=common.helpers.wrap_commands_in_shell(
'linux', task_commands),
user_identity=batchmodels.UserIdentity(auto_user=user),
resource_files=app_files,
wait_for_success=True))
common.helpers.create_pool_if_not_exist(batch_client, pool)
def pool_create():
image_reference = batchmodel.ImageReference(
publisher=config_azure['batch_pool_image_publisher'],
offer=config_azure['batch_pool_image_offer'],
sku=config_azure['batch_pool_image_sku'])
vm_config = batchmodel.VirtualMachineConfiguration(
image_reference=image_reference,
node_agent_sku_id=config_azure['batch_pool_node_agent_sku'])
vm_start_task = batchmodel.StartTask(
command_line=
'/bin/bash -c "sudo yum -y install epel-release; sudo yum -y install python36 python36-devel python36-tools; sudo python36 -m ensurepip; sudo yum -y install openmpi openmpi-devel; sudo env MPICC=/usr/lib64/openmpi/bin/mpicc pip3 install mpi4py numpy; sudo pip3 --yes uninstall azure azure-common azure-storage; sudo pip3 install azure-storage azure-batch"',
user_identity=batchmodel.UserIdentity(
auto_user=batchmodel.AutoUserSpecification(
scope=batchmodel.AutoUserScope.pool,
elevation_level=batchmodel.ElevationLevel.admin)),
wait_for_success=True)
batch_service.pool.add(pool=batchmodel.PoolAddParameter(
id=config_azure['batch_pool_name'],
vm_size=config_azure['batch_pool_vm_size'],
virtual_machine_configuration=vm_config,
target_dedicated_nodes=config_azure[
'batch_pool_target_dedicated_nodes'],
enable_inter_node_communication=True,
start_task=vm_start_task),
raw=True)
def create_pool_and_wait_for_nodes(batch_client, block_blob_client, pool_id,
vm_size, vm_count):
"""Creates an Azure Batch pool with the specified id.
:param batch_client: The batch client to use.
:type batch_client: `batchserviceclient.BatchServiceClient`
:param block_blob_client: The storage block blob client to use.
:type block_blob_client: `azure.storage.blob.BlockBlobService`
:param str pool_id: The id of the pool to create.
:param str vm_size: vm size (sku)
:param int vm_count: number of vms to allocate
:rtype: list
:return: list of `batchserviceclient.models.ComputeNode`
"""
# pick the latest supported 14.04 sku for UbuntuServer
sku_to_use, image_ref_to_use = \
common.helpers.select_latest_verified_vm_image_with_node_agent_sku(
batch_client, 'Canonical', 'UbuntuServer', '14.04')
# upload start task script
block_blob_client.create_container(_CONTAINER_NAME, fail_on_exist=False)
sas_url = common.helpers.upload_blob_and_create_sas(
block_blob_client, _CONTAINER_NAME, _STARTTASK_RESOURCE_FILE,
_STARTTASK_SHELL_SCRIPT_PATH,
datetime.datetime.utcnow() + datetime.timedelta(hours=1))
# create pool with start task
pool = batchmodels.PoolAddParameter(
id=pool_id,
enable_inter_node_communication=True,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use, node_agent_sku_id=sku_to_use),
vm_size=vm_size,
target_dedicated=vm_count,
start_task=batchmodels.StartTask(
command_line=_STARTTASK_RESOURCE_FILE,
run_elevated=True,
wait_for_success=True,
resource_files=[
batchmodels.ResourceFile(file_path=_STARTTASK_RESOURCE_FILE,
blob_source=sas_url)
]),
)
common.helpers.create_pool_if_not_exist(batch_client, pool)
# because we want all nodes to be available before any tasks are assigned
# to the pool, here we will wait for all compute nodes to reach idle
nodes = common.helpers.wait_for_all_nodes_state(
batch_client, pool,
frozenset((batchmodels.ComputeNodeState.starttaskfailed,
batchmodels.ComputeNodeState.unusable,
batchmodels.ComputeNodeState.idle)))
# ensure all node are idle
if any(node.state != batchmodels.ComputeNodeState.idle for node in nodes):
raise RuntimeError('node(s) of pool {} not in idle state'.format(
pool.id))
return nodes
def create_pool(self, pool_id):
"""
Creates a pool of compute nodes with the specified OS settings.
:param str pool_id: An ID for the new pool.
:param dict config: Configuration details.
"""
if pool_id in self.active_pools or self.pool_exists(pool_id):
return
self.logger.info("creating pool {}".format(pool_id))
pool_config = self.config['pools'][pool_id]
sku_to_use, image_ref_to_use = self.__get_vm_image_and_node_agent_sku(
pool_config)
start_vm_commands = None
if pool_config.get('create_vm_commands', None):
start_vm_commands = self.__create_commands(
pool_config['create_vm_commands'])
user = batchmodels.AutoUserSpecification(
scope=batchmodels.AutoUserScope.pool,
elevation_level=batchmodels.ElevationLevel.admin)
vm_configuration = batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use,
node_agent_sku_id=sku_to_use,
)
vm_start_task = batchmodels.StartTask(
command_line=self.__wrap_commands_in_shell('linux',
start_vm_commands),
user_identity=batchmodels.UserIdentity(auto_user=user),
wait_for_success=True)
new_pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=vm_configuration,
vm_size=pool_config['pool_vm_size'],
enable_auto_scale=True,
auto_scale_formula=pool_config['auto_scale_formula'],
auto_scale_evaluation_interval=datetime.timedelta(minutes=5),
start_task=vm_start_task,
max_tasks_per_node=pool_config['max_tasks_per_node'],
)
try:
self.batch_client.pool.add(new_pool)
except batchmodels.BatchErrorException as err:
self.__print_batch_exception(err)
raise
self.active_pools.add(pool_id)
def create_pool(self, dedicated_node_count, low_priority_node_count,
num_cores_per_node, pool_vm_size, publisher, offer, sku,
start_task, resource_files):
"""Starts pool creation on Azure
Args:
dedicated_node_count: Number of dedicated nodes in the pool
low_priority_node_count: number of low priority nodes in pool
num_cores_per_node: Number of cores per node; depends on the pool_vm_size
pool_vm_size: The type of VM to run on each node;
see https://docs.microsoft.com/en-us/azure/cloud-services/cloud-services-sizes-specs
publisher: VM image publisher (ex Canonical)
offer: VM image offer (ex UbuntuServer)
sku: VM image sku (ex 16.04)
start_task: command to run when each node joins the bool formatted as a string
resource_files: list of file references (of type azure.batch.models.ResourceFile)
from Azure Storage to download to each node
Returns:
True if creation could be started successfully. False if pool already exists.
Throws error if something else went wrong.
"""
sku_to_use, image_ref_to_use = self._select_latest_verified_vm_image_with_node_agent_sku(
publisher, offer, sku)
new_pool = batch_models.PoolAddParameter(
id=self.pool_id,
virtual_machine_configuration=batch_models.
VirtualMachineConfiguration(image_reference=image_ref_to_use,
node_agent_sku_id=sku_to_use),
vm_size=pool_vm_size,
target_dedicated_nodes=dedicated_node_count,
target_low_priority_nodes=low_priority_node_count,
start_task=self._create_start_task(start_task, resource_files),
max_tasks_per_node=num_cores_per_node,
task_scheduling_policy=batch_models.TaskSchedulingPolicy(
batch_models.ComputeNodeFillType.spread))
try:
logging.info("Attempting to create pool [{}]...".format(
self.pool_id))
self.batch_client.pool.add(new_pool)
logging.info("Pool [{}] created successfully...".format(
self.pool_id))
return True
except batch_models.BatchErrorException as e:
if e.error.code == "PoolExists":
logging.info("Pool [{}] already exists".format(self.pool_id))
return False
else:
logging.exception(
"Unknown error occurred while trying to create pool [{}]".
format(self.pool_id))
raise
def test_batch_network_configuration(self, **kwargs):
client = self.create_aad_client(**kwargs)
# Test Create Pool with Network Config
network_config = models.NetworkConfiguration(
endpoint_configuration=models.PoolEndpointConfiguration(
inbound_nat_pools=[
models.InboundNATPool(
name="TestEndpointConfig",
protocol=models.InboundEndpointProtocol.udp,
backend_port=64444,
frontend_port_range_start=60000,
frontend_port_range_end=61000,
network_security_group_rules=[
models.NetworkSecurityGroupRule(
priority=150,
access=models.NetworkSecurityGroupRuleAccess.allow,
source_address_prefix='*'
)
]
)
]
)
)
virtual_machine_config = models.VirtualMachineConfiguration(
node_agent_sku_id="batch.node.ubuntu 16.04",
image_reference=models.ImageReference(
publisher="Canonical",
offer="UbuntuServer",
sku="16.04-LTS")
)
pool = models.PoolAddParameter(
id=self.get_resource_name('batch_network_'),
target_dedicated_nodes=1,
vm_size='Standard_A1',
virtual_machine_configuration=virtual_machine_config,
network_configuration=network_config
)
client.pool.add(pool)
network_pool = client.pool.get(pool.id)
while self.is_live and network_pool.allocation_state != models.AllocationState.steady:
time.sleep(10)
network_pool = client.pool.get(pool.id)
# Test Compute Node Config
nodes = list(client.compute_node.list(pool.id))
self.assertEqual(len(nodes), 1)
self.assertIsInstance(nodes[0], models.ComputeNode)
self.assertEqual(len(nodes[0].endpoint_configuration.inbound_endpoints), 2)
self.assertEqual(nodes[0].endpoint_configuration.inbound_endpoints[0].name, 'TestEndpointConfig.0')
self.assertEqual(nodes[0].endpoint_configuration.inbound_endpoints[0].protocol.value, 'udp')
def create_pool_and_wait_for_node(batch_client, pool_id, vm_size, vm_count,
sha1_cert_tp):
"""Creates an Azure Batch pool with the specified id.
:param batch_client: The batch client to use.
:type batch_client: `batchserviceclient.BatchServiceClient`
:param str pool_id: The id of the pool to create.
:param str vm_size: vm size (sku)
:param int vm_count: number of vms to allocate
:param str sha1_cert_tp: sha1 cert thumbprint for cert ref
"""
# pick the latest supported 16.04 sku for UbuntuServer
sku_to_use, image_ref_to_use = \
common.helpers.select_latest_verified_vm_image_with_node_agent_sku(
batch_client, 'Canonical', 'UbuntuServer', '16.04')
# create start task commands
# 1. update repository
# 2. install blobxfer pre-requisites
# 3. pip install blobxfer python script
start_task_commands = [
'apt-get update',
'apt-get install -y build-essential libssl-dev libffi-dev ' +
'libpython-dev python-dev python-pip', 'pip install --upgrade blobxfer'
]
user = batchmodels.AutoUserSpecification(
scope=batchmodels.AutoUserScope.pool,
elevation_level=batchmodels.ElevationLevel.admin)
# create pool with start task and cert ref with visibility of task
pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use, node_agent_sku_id=sku_to_use),
vm_size=vm_size,
target_dedicated=vm_count,
start_task=batchmodels.StartTask(
command_line=common.helpers.wrap_commands_in_shell(
'linux', start_task_commands),
user_identity=batchmodels.UserIdentity(auto_user=user),
wait_for_success=True),
certificate_references=[
batchmodels.CertificateReference(
sha1_cert_tp,
'sha1',
visibility=[batchmodels.CertificateVisibility.task])
],
)
common.helpers.create_pool_if_not_exist(batch_client, pool)
def create_processing_pool(batch_service_client, start_task):
"""
Creates a pool of compute nodes with the specified OS settings.
:param batch_service_client: A Batch service client.
:param str start_task: task start command.
:type batch_service_client: `azure.batch.BatchServiceClient`
"""
LOGGER.info(f'Creating pool [{PROCESSING_POOL_ID}]...')
image_ref_to_use = get_image_reference()
container_registry = \
batch_models.ContainerRegistry(
registry_server=REGISTRY_SERVER,
user_name=REGISTRY_ACCOUNT_USER,
password=REGISTRY_ACCOUNT_PASSWORD)
container_conf = batch_models.ContainerConfiguration(
container_image_names=[DOCKER_CONTAINER_URL],
container_registries=[container_registry])
new_pool = batch_models.PoolAddParameter(
id=PROCESSING_POOL_ID,
virtual_machine_configuration=
batch_models.VirtualMachineConfiguration(
image_reference=image_ref_to_use,
container_configuration=container_conf,
node_agent_sku_id=VM_AGENT_SKU),
vm_size=PROCESSING_POOL_VM_SIZE,
start_task=batch_models.StartTask(
command_line=start_task,
user_identity=batch_models.UserIdentity(
auto_user=batch_models.AutoUserSpecification(
scope='pool',
elevation_level='admin'))
),
enable_auto_scale=True,
auto_scale_evaluation_interval=datetime.timedelta(
minutes=PROCESSING_POOL_SCALE_INTERVAL_MINUTES),
auto_scale_formula=PROCESSING_POOL_SCALE_FORMULA)
try:
batch_service_client.pool.add(new_pool)
LOGGER.info("Processing Pool Created")
except batch_models.BatchErrorException as err:
if 'The specified pool already exists.' in err.error.message.value:
LOGGER.info("Pool already exists...")
else:
raise
def __create_pool_and_job(self, cluster_conf, software_metadata_key: str,
start_task, VmImageModel):
"""
Create a pool and job
:param cluster_conf: the configuration object used to create the cluster
:type cluster_conf: aztk.models.ClusterConfiguration
:parm software_metadata_key: the id of the software being used on the cluster
:param start_task: the start task for the cluster
:param VmImageModel: the type of image to provision for the cluster
:param wait: wait until the cluster is ready
"""
# reuse pool_id as job_id
pool_id = cluster_conf.cluster_id
job_id = cluster_conf.cluster_id
# Get a verified node agent sku
sku_to_use, image_ref_to_use = \
helpers.select_latest_verified_vm_image_with_node_agent_sku(
VmImageModel.publisher, VmImageModel.offer, VmImageModel.sku, self.batch_client)
# Confiure the pool
pool = batch_models.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batch_models.
VirtualMachineConfiguration(image_reference=image_ref_to_use,
node_agent_sku_id=sku_to_use),
vm_size=cluster_conf.vm_size,
target_dedicated_nodes=cluster_conf.vm_count,
target_low_priority_nodes=cluster_conf.vm_low_pri_count,
start_task=start_task,
enable_inter_node_communication=True,
max_tasks_per_node=1,
metadata=[
batch_models.MetadataItem(
name=constants.AZTK_SOFTWARE_METADATA_KEY,
value=software_metadata_key),
])
# Create the pool + create user for the pool
helpers.create_pool_if_not_exist(pool, self.batch_client)
# Create job
job = batch_models.JobAddParameter(
id=job_id, pool_info=batch_models.PoolInformation(pool_id=pool_id))
# Add job to batch
self.batch_client.job.add(job)
return helpers.get_cluster(cluster_conf.cluster_id, self.batch_client)
def create_pool(config, batch_service_client):
"""
Creates a pool of compute nodes with the specified OS settings.
:param batch_service_client: A Batch service client.
:type batch_service_client: `azure.batch.BatchServiceClient`
:param str pool_id: An ID for the new pool.
:param str publisher: Marketplace image publisher
:param str offer: Marketplace image offer
:param str sku: Marketplace image sku
"""
# Create a new pool of Linux compute nodes using an Azure Virtual Machines
# Marketplace image. For more information about creating pools of Linux
# nodes, see:
# https://azure.microsoft.com/documentation/articles/batch-linux-nodes/
image_ref_to_use = models.ImageReference(
publisher="microsoft-azure-batch",
offer="ubuntu-server-container",
sku="16-04-lts",
version="latest",
)
if config.REGISTRY_USERNAME:
registry = models.ContainerRegistry(
user_name=config.REGISTRY_USERNAME,
password=config.REGISTRY_PASSWORD,
registry_server=config.REGISTRY_SERVER,
)
container_conf = models.ContainerConfiguration(
container_image_names=[config.DOCKER_CONTAINER],
container_registries=[registry],
)
else:
container_conf = models.ContainerConfiguration(
container_image_names=[config.DOCKER_CONTAINER])
new_pool = models.PoolAddParameter(
id=config.POOL_ID,
virtual_machine_configuration=models.VirtualMachineConfiguration(
image_reference=image_ref_to_use,
container_configuration=container_conf,
node_agent_sku_id="batch.node.ubuntu 16.04",
),
vm_size=config.POOL_VM_SIZE,
target_dedicated_nodes=config.POOL_NODE_COUNT,
target_low_priority_nodes=config.POOL_LOW_PRIORITY_NODE_COUNT,
)
batch_service_client.pool.add(new_pool)
def create_pool_and_wait_for_nodes(batch_client, block_blob_client, pool_id,
vm_size, vm_count):
sku_to_use, image_ref_to_use = common.helpers.select_latest_verified_vm_image_with_node_agent_sku(
batch_client, 'Canonical', 'UbuntuServer', '14.04')
block_blob_client.create_container(_CONTAINER_NAME, fail_on_exist=False)
# upload start task script
block_blob_client.create_container(_CONTAINER_NAME, fail_on_exist=False)
sas_url = common.helpers.upload_blob_and_create_sas(
block_blob_client, _CONTAINER_NAME, _STARTTASK_RESOURCE_FILE,
_STARTTASK_SHELL_SCRIPT_PATH,
datetime.datetime.utcnow() + datetime.timedelta(hours=1))
# create pool and execute starttask
pool = batchmodels.PoolAddParameter(
id=pool_id,
enable_inter_node_communication=True,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use, node_agent_sku_id=sku_to_use),
vm_size=vm_size,
target_dedicated=vm_count,
start_task=batchmodels.StartTask(
command_line=_STARTTASK_RESOURCE_FILE,
run_elevated=True,
wait_for_success=True,
resource_files=[
batchmodels.ResourceFile(file_path=_STARTTASK_RESOURCE_FILE,
blob_source=sas_url)
]),
)
common.helpers.create_pool_if_not_exist(batch_client, pool)
# because we want all nodes to be available before any tasks are assigned
# to the pool, here we will wait for all compute nodes to reach idle
nodes = common.helpers.wait_for_all_nodes_state(
batch_client, pool,
frozenset((batchmodels.ComputeNodeState.starttaskfailed,
batchmodels.ComputeNodeState.unusable,
batchmodels.ComputeNodeState.idle)))
# ensure all node are idle
if any(node.state != batchmodels.ComputeNodeState.idle for node in nodes):
raise RuntimeError('node(s) of pool {} not in idle state'.format(
pool.id))
return nodes
def create_pool(self):
from azure.batch import models as batchmodels
pool_id = self.compute_env_prefix + self.machine.name + '-' + str(
self.disk_size)
pool = self.get_pool(pool_id)
if pool is not None:
return pool_id
sku_to_use, image_ref_to_use = self.select_latest_verified_vm_image_with_node_agent_sku(
)
container_configuration = batchmodels.ContainerConfiguration(
container_image_names=[self.image])
config = batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use,
node_agent_sku_id=sku_to_use,
data_disks=[
batchmodels.DataDisk(disk_size_gb=self.disk_size, lun=1)
],
container_configuration=container_configuration,
)
pool = batchmodels.PoolAddParameter(
id=pool_id,
display_name=pool_id,
virtual_machine_configuration=config,
vm_size=self.machine.name,
)
if self.conf[utils.PLATFORM].get('low_priority', False):
pool.target_low_priority_nodes = 1
else:
pool.target_dedicated_nodes = 1
self.batch_client.pool.add(pool)
while self.get_pool(pool_id) is None:
time.sleep(1)
return pool_id
def create_pool(batch_client, container_conf, container_settings, image_ref_to_use, pool_id, sku_to_use,
vm_count, vm_size):
start_task_settings = container_settings
start_task_settings.working_directory = ContainerWorkingDirectory.container_image_default
pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use,
container_configuration=container_conf,
node_agent_sku_id=sku_to_use),
vm_size=vm_size,
max_tasks_per_node=1,
target_dedicated_nodes=vm_count,
start_task=batchmodels.StartTask(
command_line="",
wait_for_success=True,
container_settings=start_task_settings),
)
azure_helpers.create_pool_if_not_exist(batch_client, pool)
def create_commit_pool(batch_service_client):
"""
Creates a pool of compute nodes with the specified OS settings.
:param batch_service_client: A Batch service client.
:type batch_service_client: `azure.batch.BatchServiceClient`
"""
LOGGER.info(f'Creating pool [{PROCESSING_POOL_ID}]...')
image_ref_to_use = batch_models.ImageReference(
publisher='canonical',
offer='ubuntuserver',
sku='18.04-lts',
version='latest')
new_pool = batch_models.PoolAddParameter(
id=COMMIT_POOL_ID,
virtual_machine_configuration=
batch_models.VirtualMachineConfiguration(
image_reference=image_ref_to_use,
node_agent_sku_id="batch.node.ubuntu 18.04"),
vm_size=COMMIT_POOL_VM_SIZE,
start_task=batch_models.StartTask(
command_line=COMMIT_POOL_START_TASK,
user_identity=batch_models.UserIdentity(
auto_user=batch_models.AutoUserSpecification(
scope='pool',
elevation_level='admin'))
),
enable_auto_scale=True,
auto_scale_evaluation_interval=datetime.timedelta(
minutes=COMMIT_POOL_SCALE_INTERVAL_MINUTES),
auto_scale_formula=COMMIT_POOL_SCALE_FORMULA)
try:
batch_service_client.pool.add(new_pool)
LOGGER.info("Commit Pool Created")
except batch_models.BatchErrorException as err:
if 'The specified pool already exists.' in err.error.message.value:
LOGGER.info("Pool already exists...")
else:
raise
def create_pool(batch_client, block_blob_client, pool_id, vm_size, vm_count):
"""Creates an Azure Batch pool with the specified id.
:param batch_client: The batch client to use.
:type batch_client: `batchserviceclient.BatchServiceClient`
:param block_blob_client: The storage block blob client to use.
:type block_blob_client: `azure.storage.blob.BlockBlobService`
:param str pool_id: The id of the pool to create.
:param str vm_size: vm size (sku)
:param int vm_count: number of vms to allocate
"""
# pick the latest supported 14.04 sku for UbuntuServer
sku_to_use, image_ref_to_use = \
common.helpers.select_latest_verified_vm_image_with_node_agent_sku(
batch_client, 'Canonical', 'UbuntuServer', '14.04')
block_blob_client.create_container(
_CONTAINER_NAME,
fail_on_exist=False)
sas_url = common.helpers.upload_blob_and_create_sas(
block_blob_client,
_CONTAINER_NAME,
_SIMPLE_TASK_NAME,
_SIMPLE_TASK_PATH,
datetime.datetime.utcnow() + datetime.timedelta(hours=1))
pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use,
node_agent_sku_id=sku_to_use),
vm_size=vm_size,
target_dedicated=vm_count,
start_task=batchmodels.StartTask(
command_line="python " + _SIMPLE_TASK_NAME,
resource_files=[batchmodels.ResourceFile(
file_path=_SIMPLE_TASK_NAME,
blob_source=sas_url)]))
common.helpers.create_pool_if_not_exist(batch_client, pool)
def create_pool(batch_client, block_blob_client, pool_id, vm_size, vm_count):
"""Creates an Azure Batch pool with the specified id.
:param batch_client: The batch client to use.
:type batch_client: `batchserviceclient.BatchServiceClient`
:param block_blob_client: The storage block blob client to use.
:type block_blob_client: `azure.storage.blob.BlockBlobService`
:param str pool_id: The id of the pool to create.
:param str vm_size: vm size (sku)
:param int vm_count: number of vms to allocate
"""
# pick the latest supported 16.04 sku for UbuntuServer
sku_to_use, image_ref_to_use = \
common.helpers.select_latest_verified_vm_image_with_node_agent_sku(
batch_client, 'Canonical', 'UbuntuServer', '18.04')
block_blob_client.create_container(_CONTAINER_NAME, fail_on_exist=False)
sas_url = common.helpers.upload_blob_and_create_sas(
block_blob_client, _CONTAINER_NAME, _SIMPLE_TASK_NAME,
_SIMPLE_TASK_PATH, _EXPIRY_TIME)
start_tasks = []
pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use, node_agent_sku_id=sku_to_use),
vm_size=vm_size,
target_dedicated_nodes=vm_count,
start_task=batchmodels.StartTask(
command_line=
'/bin/bash -c \"sudo apt-get -y update && export DEBIAN_FRONTEND=noninteractive && sudo apt-get install -y python3-pip && sudo pip3 install numpy statsmodels pmdarima\"',
wait_for_success=True,
user_identity=batchmodels.UserIdentity(
auto_user=batchmodels.AutoUserSpecification(
scope=batchmodels.AutoUserScope.pool,
elevation_level=batchmodels.ElevationLevel.admin)),
))
common.helpers.create_pool_if_not_exist(batch_client, pool)
def _create_pool(self):
"""
Creates a pool of compute nodes with the specified OS settings.
:param batch_service_client: A Batch service client.
:type batch_service_client: `azure.batch.BatchServiceClient`
:param str pool_id: An ID for the new pool.
:param str publisher: Marketplace image publisher
:param str offer: Marketplace image offer
:param str sku: Marketplace image sku
"""
if self.config.REGISTRY_SERVER:
print("Using a private registry")
registry = models.ContainerRegistry(
user_name=self.config.REGISTRY_USERNAME,
password=self.config.REGISTRY_PASSWORD,
registry_server=self.config.REGISTRY_SERVER,
)
container_conf = models.ContainerConfiguration(
container_image_names=[self.config.DOCKER_IMAGE],
container_registries=[registry],
)
else:
container_conf = models.ContainerConfiguration(
container_image_names=[self.config.DOCKER_IMAGE])
new_pool = models.PoolAddParameter(
id=self.config.POOL_ID,
virtual_machine_configuration=models.VirtualMachineConfiguration(
image_reference=_IMAGE_REF,
container_configuration=container_conf,
node_agent_sku_id="batch.node.ubuntu 16.04",
),
vm_size=self.config.POOL_VM_SIZE,
target_dedicated_nodes=self.config.POOL_NODE_COUNT,
target_low_priority_nodes=self.config.POOL_LOW_PRIORITY_NODE_COUNT,
)
# Create the pool
self.batch_client.pool.add(new_pool)
def create_pool(batch_client, block_blob_client, pool_id, vm_size, vm_count):
block_blob_client.create_container(CONTAINER_NAME, fail_on_exist=False)
sku_to_use, image_ref_to_use = select_latest_vm_image_with_node_agent_sku(
batch_client, 'Canonical', 'UbuntuServer', '18.04')
sas_url = upload_blob_and_create_sas(
block_blob_client, CONTAINER_NAME, TASK_NAME, TASK_PATH,
datetime.datetime.utcnow() + datetime.timedelta(hours=1))
pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use, node_agent_sku_id=sku_to_use),
vm_size=vm_size,
target_dedicated_nodes=vm_count,
start_task=batchmodels.StartTask(command_line="python3 " + TASK_NAME,
resource_files=[
batchmodels.ResourceFile(
file_path=TASK_NAME,
blob_source=sas_url)
]))
create_pool_if_not_exist(batch_client, pool)
def create_pool_and_wait_for_node(batch_client, pool_id, vm_size, vm_count,
sha1_cert_tp):
"""Creates an Azure Batch pool with the specified id.
:param batch_client: The batch client to use.
:type batch_client: `batchserviceclient.BatchServiceClient`
:param str pool_id: The id of the pool to create.
:param str vm_size: vm size (sku)
:param int vm_count: number of vms to allocate
:param str sha1_cert_tp: sha1 cert thumbprint for cert ref
"""
# pick the latest supported 16.04 sku for UbuntuServer
sku_to_use, image_ref_to_use = \
common.helpers.select_latest_verified_vm_image_with_node_agent_sku(
batch_client, 'Canonical', 'UbuntuServer', '16.04')
# create start task commands
# 1. update repository
# 2. install blobxfer pre-requisites
# 3. pip install blobxfer python script
start_task_commands = [
'apt-get update',
'apt-get install -y build-essential libssl-dev libffi-dev ' +
'libpython-dev python-dev python-pip', 'pip install --upgrade blobxfer'
]
user = batchmodels.AutoUserSpecification(
scope=batchmodels.AutoUserScope.pool,
elevation_level=batchmodels.ElevationLevel.admin)
# create pool with start task and cert ref with visibility of task
pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=batchmodels.VirtualMachineConfiguration(
image_reference=image_ref_to_use, node_agent_sku_id=sku_to_use),
vm_size=vm_size,
target_dedicated_nodes=vm_count,
start_task=batchmodels.StartTask(
command_line=common.helpers.wrap_commands_in_shell(
'linux', start_task_commands),
user_identity=batchmodels.UserIdentity(auto_user=user),
wait_for_success=True),
certificate_references=[
batchmodels.CertificateReference(
thumbprint=sha1_cert_tp,
thumbprint_algorithm='sha1',
visibility=[batchmodels.CertificateVisibility.task])
],
)
common.helpers.create_pool_if_not_exist(batch_client, pool)
# because we want all nodes to be available before any tasks are assigned
# to the pool, here we will wait for all compute nodes to reach idle
nodes = common.helpers.wait_for_all_nodes_state(
batch_client, pool,
frozenset((batchmodels.ComputeNodeState.start_task_failed,
batchmodels.ComputeNodeState.unusable,
batchmodels.ComputeNodeState.idle)))
# ensure all node are idle
if any(node.state != batchmodels.ComputeNodeState.idle for node in nodes):
raise RuntimeError(
'node(s) of pool {} not in idle state'.format(pool_id))
def test_batch_update_pools(self, **kwargs):
client = self.create_sharedkey_client(**kwargs)
# Test Create Paas Pool
test_paas_pool = models.PoolAddParameter(
id=self.get_resource_name('batch_paas_'),
vm_size='small',
cloud_service_configuration=models.CloudServiceConfiguration(
os_family='5'
),
start_task=models.StartTask(
command_line="cmd.exe /c \"echo hello world\"",
resource_files=[models.ResourceFile('https://blobsource.com', 'filename.txt')],
environment_settings=[models.EnvironmentSetting('ENV_VAR', 'env_value')],
user_identity=models.UserIdentity(
auto_user=models.AutoUserSpecification(
elevation_level=models.ElevationLevel.admin
)
)
)
)
response = client.pool.add(test_paas_pool)
self.assertIsNone(response)
# Test Upgrade Pool OS
self.assertBatchError(
"PoolVersionEqualsUpgradeVersion",
client.pool.upgrade_os,
test_paas_pool.id,
"*"
)
# Test Update Pool Parameters
params = models.PoolUpdatePropertiesParameter([], [], [models.MetadataItem('foo', 'bar')])
response = client.pool.update_properties(test_paas_pool.id, params)
self.assertIsNone(response)
# Test Patch Pool Parameters
params = models.PoolPatchParameter(metadata=[models.MetadataItem('foo2', 'bar2')])
response = client.pool.patch(test_paas_pool.id, params)
self.assertIsNone(response)
# Test Pool Exists
response = client.pool.exists(test_paas_pool.id)
self.assertTrue(response)
# Test Get Pool
pool = client.pool.get(test_paas_pool.id)
self.assertIsInstance(pool, models.CloudPool)
self.assertEqual(pool.id, test_paas_pool.id)
self.assertEqual(pool.state, models.PoolState.active)
self.assertEqual(pool.allocation_state, models.AllocationState.steady)
self.assertEqual(pool.cloud_service_configuration.os_family, '5')
self.assertEqual(pool.vm_size, 'small')
self.assertIsNone(pool.start_task)
self.assertEqual(pool.metadata[0].name, 'foo2')
self.assertEqual(pool.metadata[0].value, 'bar2')
# Test Get Pool with OData Clauses
options = models.PoolGetOptions(select='id,state', expand='stats')
pool = client.pool.get(test_paas_pool.id, options)
self.assertIsInstance(pool, models.CloudPool)
self.assertEqual(pool.id, test_paas_pool.id)
self.assertEqual(pool.state, models.PoolState.active)
self.assertIsNone(pool.allocation_state)
self.assertIsNone(pool.vm_size)
# Test Delete Pool
response = client.pool.delete(test_paas_pool.id)
self.assertIsNone(response)
def test_batch_create_pools(self, **kwargs):
client = self.create_sharedkey_client(**kwargs)
# Test List Node Agent SKUs
response = client.account.list_node_agent_skus()
response = list(response)
self.assertTrue(len(response) > 1)
self.assertEqual(response[-1].id, "batch.node.windows amd64")
self.assertEqual(response[-1].os_type.value, "windows")
self.assertTrue(len(response[-1].verified_image_references) > 1)
# Test Create Iaas Pool
users = [
models.UserAccount('test-user-1', 'kt#_gahr!@aGERDXA'),
models.UserAccount('test-user-2', 'kt#_gahr!@aGERDXA', models.ElevationLevel.admin)
]
test_iaas_pool = models.PoolAddParameter(
id=self.get_resource_name('batch_iaas_'),
vm_size='Standard_A1',
virtual_machine_configuration=models.VirtualMachineConfiguration(
image_reference=models.ImageReference(
publisher='MicrosoftWindowsServer',
offer='WindowsServer',
sku='2016-Datacenter-smalldisk'
),
node_agent_sku_id='batch.node.windows amd64',
windows_configuration=models.WindowsConfiguration(True)),
task_scheduling_policy=models.TaskSchedulingPolicy(models.ComputeNodeFillType.pack),
user_accounts=users
)
response = client.pool.add(test_iaas_pool)
self.assertIsNone(response)
# Test list pool node counnt
counts = list(client.account.list_pool_node_counts())
self.assertIsNotNone(counts)
self.assertEqual(len(counts), 1)
self.assertEqual(counts[0].pool_id, test_iaas_pool.id)
self.assertIsNotNone(counts[0].dedicated)
self.assertEqual(counts[0].dedicated.total, 0)
self.assertEqual(counts[0].low_priority.total, 0)
# Test Create Pool with Network Configuration
network_config = models.NetworkConfiguration('/subscriptions/00000000-0000-0000-0000-000000000000'
'/resourceGroups/test'
'/providers/Microsoft.Network'
'/virtualNetworks/vnet1'
'/subnets/subnet1')
test_network_pool = models.PoolAddParameter(
id=self.get_resource_name('batch_network_'),
vm_size='Standard_A1',
network_configuration=network_config,
virtual_machine_configuration=models.VirtualMachineConfiguration(
image_reference=models.ImageReference(
publisher='Canonical',
offer='UbuntuServer',
sku='16.04-LTS'
),
node_agent_sku_id='batch.node.ubuntu 16.04')
)
self.assertBatchError('InvalidPropertyValue', client.pool.add, test_network_pool, models.PoolAddOptions(timeout=45))
# Test Create Pool with Custom Image
test_image_pool = models.PoolAddParameter(
id=self.get_resource_name('batch_image_'),
vm_size='Standard_A1',
virtual_machine_configuration=models.VirtualMachineConfiguration(
image_reference=models.ImageReference(
virtual_machine_image_id="/subscriptions/00000000-0000-0000-0000-000000000000"
"/resourceGroups/test"
"/providers/Microsoft.Compute"
"/images/FakeImage"
),
node_agent_sku_id='batch.node.ubuntu 16.04'
)
)
self.assertBatchError('InvalidPropertyValue', client.pool.add, test_image_pool, models.PoolAddOptions(timeout=45))
# Test Create Pool with OSDisk
os_disk = models.OSDisk(caching=models.CachingType.read_write)
test_osdisk_pool = models.PoolAddParameter(
id=self.get_resource_name('batch_osdisk_'),
vm_size='Standard_A1',
virtual_machine_configuration=models.VirtualMachineConfiguration(
image_reference=models.ImageReference(
publisher='Canonical',
offer='UbuntuServer',
sku='16.04-LTS'
),
node_agent_sku_id='batch.node.ubuntu 16.04',
os_disk=os_disk)
)
response = client.pool.add(test_osdisk_pool)
self.assertIsNone(response)
osdisk_pool = client.pool.get(test_osdisk_pool.id)
self.assertEqual(osdisk_pool.virtual_machine_configuration.os_disk.caching, os_disk.caching)
# Test Create Pool with Data Disk
data_disk = models.DataDisk(lun=1, disk_size_gb=50)
test_disk_pool = models.PoolAddParameter(
id=self.get_resource_name('batch_disk_'),
vm_size='Standard_A1',
virtual_machine_configuration=models.VirtualMachineConfiguration(
image_reference=models.ImageReference(
publisher='Canonical',
offer='UbuntuServer',
sku='16.04-LTS'
),
node_agent_sku_id='batch.node.ubuntu 16.04',
data_disks=[data_disk])
)
response = client.pool.add(test_disk_pool)
self.assertIsNone(response)
disk_pool = client.pool.get(test_disk_pool.id)
self.assertEqual(disk_pool.virtual_machine_configuration.data_disks[0].lun, 1)
self.assertEqual(disk_pool.virtual_machine_configuration.data_disks[0].disk_size_gb, 50)
# Test Create Pool with Application Licenses
test_app_pool = models.PoolAddParameter(
id=self.get_resource_name('batch_app_'),
vm_size='Standard_A1',
application_licenses=["maya"],
virtual_machine_configuration=models.VirtualMachineConfiguration(
image_reference=models.ImageReference(
publisher='Canonical',
offer='UbuntuServer',
sku='16.04-LTS'
),
node_agent_sku_id='batch.node.ubuntu 16.04',
data_disks=[data_disk])
)
response = client.pool.add(test_app_pool)
self.assertIsNone(response)
app_pool = client.pool.get(test_app_pool.id)
self.assertEqual(app_pool.application_licenses[0], "maya")
# Test List Pools without Filters
pools = list(client.pool.list())
self.assertTrue(len(pools) > 1)
# Test List Pools with Maximum
options = models.PoolListOptions(max_results=1)
pools = client.pool.list(options)
pools.next()
self.assertEqual(len(pools.current_page), 1)
# Test List Pools with Filter
options = models.PoolListOptions(
filter='startswith(id,\'batch_app_\')',
select='id,state',
expand='stats')
pools = list(client.pool.list(options))
self.assertEqual(len(pools), 1)
def configure_pool(
self,
pool_id: str,
vm_size: Optional[str] = None,
vm_publisher: Optional[str] = None,
vm_offer: Optional[str] = None,
sku_starts_with: Optional[str] = None,
vm_sku: Optional[str] = None,
vm_version: Optional[str] = None,
vm_node_agent_sku_id: Optional[str] = None,
os_family: Optional[str] = None,
os_version: Optional[str] = None,
display_name: Optional[str] = None,
target_dedicated_nodes: Optional[int] = None,
use_latest_image_and_sku: bool = False,
**kwargs,
) -> PoolAddParameter:
"""
Configures a pool
:param pool_id: A string that uniquely identifies the Pool within the Account
:param vm_size: The size of virtual machines in the Pool.
:param display_name: The display name for the Pool
:param target_dedicated_nodes: The desired number of dedicated Compute Nodes in the Pool.
:param use_latest_image_and_sku: Whether to use the latest verified vm image and sku
:param vm_publisher: The publisher of the Azure Virtual Machines Marketplace Image.
For example, Canonical or MicrosoftWindowsServer.
:param vm_offer: The offer type of the Azure Virtual Machines Marketplace Image.
For example, UbuntuServer or WindowsServer.
:param sku_starts_with: The start name of the sku to search
:param vm_sku: The name of the virtual machine sku to use
:param vm_version: The version of the virtual machine
:param vm_version: str
:param vm_node_agent_sku_id: The node agent sku id of the virtual machine
:param os_family: The Azure Guest OS family to be installed on the virtual machines in the Pool.
:param os_version: The OS family version
"""
if use_latest_image_and_sku:
self.log.info(
'Using latest verified virtual machine image with node agent sku'
)
sku_to_use, image_ref_to_use = self._get_latest_verified_image_vm_and_sku(
publisher=vm_publisher,
offer=vm_offer,
sku_starts_with=sku_starts_with)
pool = batch_models.PoolAddParameter(
id=pool_id,
vm_size=vm_size,
display_name=display_name,
virtual_machine_configuration=batch_models.
VirtualMachineConfiguration(image_reference=image_ref_to_use,
node_agent_sku_id=sku_to_use),
target_dedicated_nodes=target_dedicated_nodes,
**kwargs,
)
elif os_family:
self.log.info(
'Using cloud service configuration to create pool, virtual machine configuration ignored'
)
pool = batch_models.PoolAddParameter(
id=pool_id,
vm_size=vm_size,
display_name=display_name,
cloud_service_configuration=batch_models.
CloudServiceConfiguration(os_family=os_family,
os_version=os_version),
target_dedicated_nodes=target_dedicated_nodes,
**kwargs,
)
else:
self.log.info(
'Using virtual machine configuration to create a pool')
pool = batch_models.PoolAddParameter(
id=pool_id,
vm_size=vm_size,
display_name=display_name,
virtual_machine_configuration=batch_models.
VirtualMachineConfiguration(
image_reference=batch_models.ImageReference(
publisher=vm_publisher,
offer=vm_offer,
sku=vm_sku,
version=vm_version,
),
node_agent_sku_id=vm_node_agent_sku_id,
),
target_dedicated_nodes=target_dedicated_nodes,
**kwargs,
)
return pool
def check_or_create_pool(self, id=None):
if id is None:
id = self.config.get('POOL', 'id')
self.pool_id = id
if self.client.pool.exists(id):
found_job = False
# Update the Job ID here
for job in self.client.job.list():
if job.pool_info.pool_id == self.pool_id:
self.job_id = job.id
found_job = True
break
if not found_job:
self.start_mc_server_job_pool(
) # Restart Jobs for this pool - this is necessary!
return self.client.pool.get(id)
api_port = self.config.get('POOL', 'api_port')
min_count = self.config.get('POOL', 'mincount')
image_reference = batchmodels.ImageReference(
virtual_machine_image_id=
"/subscriptions/889566d5-6e5d-4d31-a82d-b60603b3e50b/resourceGroups/polycraft-game/providers/Microsoft.Compute/galleries/polycraftImgGallery/images/polycraftBestGameServerV1/versions/1.0.0"
)
vmc = batchmodels.VirtualMachineConfiguration(
image_reference=image_reference,
node_agent_sku_id="batch.node.ubuntu 18.04")
users = [
batchmodels.UserAccount(
name='azureuser',
password='******',
elevation_level=batchmodels.ElevationLevel.admin),
# batchmodels.UserAccount(
# name='pool-nonadmin',
# password='******',
# elevation_level=batchmodels.ElevationLevel.non_admin)
]
# Thank you Ask Ubuntu https://askubuntu.com/a/373478
wait_for_locks = 'while sudo fuser /var/lib/dpkg/lock /var/lib/apt/lists/lock /var/cache/apt/archives/lock /var/lib/dpkg/lock-frontend >/dev/null 2>&1; do echo "Waiting for release of apt locks"; sleep 2; done; '
# NOTE: Always use DOUBLE QUOTES within commands as azure prepends the entire string with a single quote.
start_task = batchmodels.StartTask(
command_line=helpers.wrap_commands_in_shell(
'linux',
[
'whoami',
'printenv',
'usermod -aG sudo azureuser',
'sudo systemctl disable --now apt-daily.timer',
'sudo systemctl disable --now apt-daily-upgrade.timer',
'sudo systemctl daemon-reload',
'cd /home/polycraft',
'chmod -R 777 *',
'rm /home/polycraft/oxygen/mods/*.jar',
'cd /home/polycraft/oxygen/',
'echo "[DEBUG] removing helium..."',
'ls -l',
f'sudo rm -rf /home/polycraft/oxygen/{self.config.get("SERVER","worldName")}',
'sudo rm -f *.zip',
'echo "[DEBUG] removed helium?"',
'ls -l',
# Stop the crontabs from running
'sudo rm /var/spool/cron/crontabs/*',
# Taken from: https://stackoverflow.com/questions/45269225/ansible-playbook-fails-to-lock-apt/51919678#51919678
'sudo systemd-run --property="After=apt-daily.service apt-daily-upgrade.service" --wait /bin/true',
'sudo apt-get -y purge unattended-upgrades',
'sudo apt-get -y update',
wait_for_locks +
'sudo apt-get install software-properties-common -y',
# 'while fuser /var/lib/dpkg/lock >/dev/null 2>&1; do sleep 1; done; sudo apt-add-repository universe',
wait_for_locks + 'sudo apt-add-repository universe',
# Mount the Polycraft Game FileShare
wait_for_locks +
'sudo apt-get install cifs-utils -y && sudo mkdir -p /mnt/PolycraftGame/',
f'mount -t cifs //polycraftbestbatch.file.core.windows.net/best-batch-round-1-test /mnt/PolycraftGame -o vers=3.0,username={self.credentials.get("Storage", "storageaccountname")},password={self.credentials.get("Storage", "storageaccountkey")},dir_mode=0777,file_mode=0777,serverino && ls /mnt/PolycraftGame',
# Copy the default world file to the right folder
f'cp /mnt/PolycraftGame/{self.config.get("SERVER","fileShareFolder")}/worlds/{self.config.get("SERVER","worldZipName")}.tar.gz /home/polycraft/oxygen/',
'cd /home/polycraft/oxygen/',
# 'sudo rm -r helium',
f'gzip -d /home/polycraft/oxygen/{self.config.get("SERVER","worldZipName")}.tar.gz',
'echo "[DEBUG] extracting the tar"',
'ls -l',
f'sudo tar -xf {self.config.get("SERVER","worldZipName")}.tar',
'echo "[DEBUG] extracted the tar"',
'ls -l',
# 'sudo mv helium-backup-0924 helium',
f'sudo mv helium {self.config.get("SERVER","worldName")}', # TODO Remove this once we finalize the server name?
f'chmod -R 777 {self.config.get("SERVER","worldName")}/', # NOTE: The folder inside here is called helium!
'echo "[DEBUG] Adjusted permissions for helium?"',
'ls -l',
]),
wait_for_success=True,
# user_accounts=users,
user_identity=batchmodels.UserIdentity(
# user_name='azureuser',
auto_user=batchmodels.AutoUserSpecification(
scope=batchmodels.AutoUserScope.pool,
elevation_level=batchmodels.ElevationLevel.admin)
# ),
),
)
net_config = batchmodels.NetworkConfiguration(
# subnet_id="/subscriptions/889566d5-6e5d-4d31-a82d-b60603b3e50b/resourceGroups/vnet-eastus-azurebatch/providers/Microsoft.Network/virtualNetworks/vnet-eastus-azurebatch/subnets/main-batch-subnet",
endpoint_configuration=batchmodels.
PoolEndpointConfiguration(inbound_nat_pools=[
batchmodels.InboundNATPool(
name='minecraftServer',
protocol='tcp',
backend_port=25565,
frontend_port_range_start=44000,
frontend_port_range_end=44099,
network_security_group_rules=[
batchmodels.NetworkSecurityGroupRule(
priority=199,
access='allow',
source_address_prefix='*'),
]),
batchmodels.InboundNATPool(
name='api_port',
protocol='tcp',
backend_port=int(api_port)
if api_port and api_port.isdecimal() else 9007,
frontend_port_range_start=44500,
frontend_port_range_end=44599,
network_security_group_rules=[
# batchmodels.NetworkSecurityGroupRule(
# priority=170,
# access='allow',
# source_address_prefix='192.168.1.0/24' # TODO: is this the right subnet?
# ),
batchmodels.NetworkSecurityGroupRule(
priority=198,
access='allow', # 'deny'
source_address_prefix=
'*' # TODO: only allow access to the right ports
)
]),
]))
pool = batchmodels.PoolAddParameter(
id=id,
vm_size=self.config.get('POOL', 'vm_size'),
target_dedicated_nodes=int(min_count)
if min_count and min_count.isdecimal() else 1,
virtual_machine_configuration=vmc,
start_task=start_task,
user_accounts=users,
network_configuration=net_config)
helpers.create_pool_if_not_exist(self.client, pool)
self.start_mc_server_job_pool(pool.target_dedicated_nodes)
def create_pool_with_containers(batch_service_client, pool_id, resource_files,
publisher, offer, sku):
"""
Creates a pool of compute nodes with the specified OS settings.
:param batch_service_client: A Batch service client.
:type batch_service_client: `azure.batch.BatchServiceClient`
:param str pool_id: An ID for the new pool.
:param list resource_files: A collection of resource files for the pool's
start task.
:param str publisher: Marketplace image publisher
:param str offer: Marketplace image offer
:param str sku: Marketplace image sku
"""
print('Creating pool [{}]...'.format(pool_id))
# Create a new pool of Linux compute nodes using an Azure Virtual Machines
# Marketplace image. For more information about creating pools of Linux
# nodes, see:
# https://azure.microsoft.com/documentation/articles/batch-linux-nodes/
# Specify the commands for the pool's start task. The start task is run
# on each node as it joins the pool, and when it's rebooted or re-imaged.
# We use the start task to prep the node for running our task script.
task_commands = [
# Copy the python_tutorial_task.py script to the "shared" directory
# that all tasks that run on the node have access to. Note that
# we are using the -p flag with cp to preserve the file uid/gid,
# otherwise since this start task is run as an admin, it would not
# be accessible by tasks run as a non-admin user.
#'wget https://packages.microsoft.com/config/ubuntu/16.04/packages-microsoft-prod.deb',
#'sudo dpkg -i packages-microsoft-prod.deb',
#'wget -O azcopy.tar.gz https://aka.ms/downloadazcopylinux64',
#'tar -xf azcopy.tar.gz',
#'sudo ./install.sh',
#'wget https://repo.anaconda.com/archive/Anaconda3-5.1.0-Linux-x86_64.sh -O ~/conda.sh',
#'bash ~/conda.sh -b -p $AZ_BATCH_NODE_SHARED_DIR/conda',
#'export PATH="$AZ_BATCH_NODE_SHARED_DIR/conda/bin:$PATH"',
#'sudo apt-get -y update',
#'sudo apt-get -y install azcopy',
'cp -p {} $AZ_BATCH_NODE_SHARED_DIR'.format(_TUTORIAL_TASK_FILE),
#'cp -p {} $AZ_BATCH_NODE_SHARED_DIR'.format(_ENV_YML_FILE),
'azcopy --source https://{0}.blob.core.windows.net/model/ghanamines.h5 --destination $AZ_BATCH_NODE_SHARED_DIR/ghanamines.h5 --source-key {1}'
.format(_STORAGE_ACCOUNT_NAME, _STORAGE_ACCOUNT_KEY),
#'sudo $AZ_BATCH_NODE_SHARED_DIR/conda/bin/conda env create -f {}'.format(_ENV_YML_FILE)
]
# Get the node agent SKU and image reference for the virtual machine
# configuration.
# For more information about the virtual machine configuration, see:
# https://azure.microsoft.com/documentation/articles/batch-linux-nodes/
sku_to_use, image_ref_to_use = \
common_helpers.select_latest_verified_vm_image_with_node_agent_sku(
batch_service_client, publisher, offer, sku)
user = batchmodels.AutoUserSpecification(
scope=batchmodels.AutoUserScope.pool,
elevation_level=batchmodels.ElevationLevel.admin)
container_reg = batchmodels.ContainerRegistry(user_name=CLIENT_ID,
password=SECRET,
registry_server=_ACR_URL)
container_cfg = batchmodels.ContainerConfiguration(
container_image_names=[_ACR_IMG_NAME],
container_registries=[container_reg])
my_img_ref = batchmodels.ImageReference(
virtual_machine_image_id=_CUSTOM_VM_IMG_ID)
vm_cfg = batchmodels.VirtualMachineConfiguration(
image_reference=my_img_ref,
node_agent_sku_id=
sku_to_use, #'batch.node.ubuntu 16.04', ##verificare che l'immagine ghanaimg abbia gpu
container_configuration=container_cfg)
task_containersettings = batchmodels.TaskContainerSettings(
image_name=_ACR_IMG_NAME)
new_pool = batchmodels.PoolAddParameter(
id=pool_id,
virtual_machine_configuration=vm_cfg,
vm_size=_POOL_VM_SIZE,
target_dedicated_nodes=_POOL_NODE_COUNT,
target_low_priority_nodes=1,
start_task=batch.models.StartTask(
command_line=common_helpers.wrap_commands_in_shell(
'linux', task_commands),
user_identity=batchmodels.UserIdentity(auto_user=user),
wait_for_success=True,
resource_files=resource_files,
container_settings=task_containersettings))
try:
batch_service_client.pool.add(new_pool)
except batchmodels.batch_error.BatchErrorException as err:
print_batch_exception(err)
raise
