def __init__(self, dashboard_head):
super().__init__(dashboard_head)
self.cluster_metadata = ray_usage_lib.get_cluster_metadata(
ray.experimental.internal_kv.internal_kv_get_gcs_client(),
num_retries=20,
)
self.cluster_config_to_report = ray_usage_lib.get_cluster_config_to_report(
os.path.expanduser("~/ray_bootstrap_config.yaml"))
self.session_dir = dashboard_head.session_dir
self.client = ray_usage_lib.UsageReportClient()
# The total number of report succeeded.
self.total_success = 0
# The total number of report failed.
self.total_failed = 0
# The seq number of report. It increments whenever a new report is sent.
self.seq_no = 0
def test_usage_lib_report_data(
monkeypatch, ray_start_cluster, tmp_path, reset_lib_usage
):
with monkeypatch.context() as m:
m.setenv("RAY_USAGE_STATS_ENABLED", "1")
m.setenv("RAY_USAGE_STATS_REPORT_URL", "http://127.0.0.1:8000")
cluster = ray_start_cluster
cluster.add_node(num_cpus=0)
# Runtime env is required to run this test in minimal installation test.
ray.init(address=cluster.address, runtime_env={"pip": ["ray[serve]"]})
"""
Make sure the generated data is following the schema.
"""
cluster_metadata = ray_usage_lib.get_cluster_metadata(
ray.experimental.internal_kv.internal_kv_get_gcs_client(), num_retries=20
)
cluster_config_file_path = tmp_path / "ray_bootstrap_config.yaml"
cluster_config_file_path.write_text(
"""
cluster_name: minimal
max_workers: 1
provider:
type: aws
region: us-west-2
availability_zone: us-west-2a
"""
)
cluster_config_to_report = ray_usage_lib.get_cluster_config_to_report(
cluster_config_file_path
)
d = ray_usage_lib.generate_report_data(
cluster_metadata, cluster_config_to_report, 2, 2, 2
)
validate(instance=asdict(d), schema=schema)
"""
Make sure writing to a file works as expected
"""
client = ray_usage_lib.UsageReportClient()
temp_dir = Path(tmp_path)
client.write_usage_data(d, temp_dir)
wait_for_condition(lambda: file_exists(temp_dir))
"""
Make sure report usage data works as expected
"""
@ray.remote(num_cpus=0)
class ServeInitator:
def __init__(self):
# Start the ray serve server to verify requests are sent
# to the right place.
from ray import serve
serve.start()
@serve.deployment(ray_actor_options={"num_cpus": 0})
async def usage(request):
body = await request.json()
if body == asdict(d):
return True
else:
return False
usage.deploy()
def ready(self):
pass
# We need to start a serve with runtime env to make this test
# work with minimal installation.
s = ServeInitator.remote()
ray.get(s.ready.remote())
# Query our endpoint over HTTP.
r = client.report_usage_data("http://127.0.0.1:8000/usage", d)
r.raise_for_status()
assert json.loads(r.text) is True
def test_usage_lib_get_cluster_config_to_report(monkeypatch, tmp_path, reset_lib_usage):
cluster_config_file_path = tmp_path / "ray_bootstrap_config.yaml"
""" Test minimal cluster config"""
cluster_config_file_path.write_text(
"""
cluster_name: minimal
max_workers: 1
provider:
type: aws
region: us-west-2
availability_zone: us-west-2a
"""
)
cluster_config_to_report = ray_usage_lib.get_cluster_config_to_report(
cluster_config_file_path
)
assert cluster_config_to_report.cloud_provider == "aws"
assert cluster_config_to_report.min_workers is None
assert cluster_config_to_report.max_workers == 1
assert cluster_config_to_report.head_node_instance_type is None
assert cluster_config_to_report.worker_node_instance_types is None
cluster_config_file_path.write_text(
"""
cluster_name: full
min_workers: 1
provider:
type: gcp
head_node_type: head_node
available_node_types:
head_node:
node_config:
InstanceType: m5.large
min_workers: 0
max_workers: 0
aws_worker_node:
node_config:
InstanceType: m3.large
min_workers: 0
max_workers: 0
azure_worker_node:
node_config:
azure_arm_parameters:
vmSize: Standard_D2s_v3
gcp_worker_node:
node_config:
machineType: n1-standard-2
"""
)
cluster_config_to_report = ray_usage_lib.get_cluster_config_to_report(
cluster_config_file_path
)
assert cluster_config_to_report.cloud_provider == "gcp"
assert cluster_config_to_report.min_workers == 1
assert cluster_config_to_report.max_workers is None
assert cluster_config_to_report.head_node_instance_type == "m5.large"
assert cluster_config_to_report.worker_node_instance_types == list(
{"m3.large", "Standard_D2s_v3", "n1-standard-2"}
)
cluster_config_file_path.write_text(
"""
cluster_name: full
head_node_type: head_node
available_node_types:
worker_node_1:
node_config:
ImageId: xyz
worker_node_2:
resources: {}
worker_node_3:
node_config:
InstanceType: m5.large
"""
)
cluster_config_to_report = ray_usage_lib.get_cluster_config_to_report(
cluster_config_file_path
)
assert cluster_config_to_report.cloud_provider is None
assert cluster_config_to_report.min_workers is None
assert cluster_config_to_report.max_workers is None
assert cluster_config_to_report.head_node_instance_type is None
assert cluster_config_to_report.worker_node_instance_types == ["m5.large"]
cluster_config_file_path.write_text("[invalid")
cluster_config_to_report = ray_usage_lib.get_cluster_config_to_report(
cluster_config_file_path
)
assert cluster_config_to_report == ClusterConfigToReport()
cluster_config_to_report = ray_usage_lib.get_cluster_config_to_report(
tmp_path / "does_not_exist.yaml"
)
assert cluster_config_to_report == ClusterConfigToReport()
monkeypatch.setenv("KUBERNETES_SERVICE_HOST", "localhost")
cluster_config_to_report = ray_usage_lib.get_cluster_config_to_report(
tmp_path / "does_not_exist.yaml"
)
assert cluster_config_to_report.cloud_provider == "kubernetes"
assert cluster_config_to_report.min_workers is None
assert cluster_config_to_report.max_workers is None
assert cluster_config_to_report.head_node_instance_type is None
assert cluster_config_to_report.worker_node_instance_types is None
