def test_deploy_cli_creates_sagemaker_transform_job_and_s3_resources_with_expected_names_from_s3(
pretrained_model, sagemaker_client):
local_model_path = _download_artifact_from_uri(pretrained_model.model_uri)
artifact_path = "model"
region_name = sagemaker_client.meta.region_name
default_bucket = mfs._get_default_s3_bucket(region_name)
s3_artifact_repo = S3ArtifactRepository("s3://{}".format(default_bucket))
s3_artifact_repo.log_artifacts(local_model_path,
artifact_path=artifact_path)
model_s3_uri = "s3://{bucket_name}/{artifact_path}".format(
bucket_name=default_bucket, artifact_path=pretrained_model.model_path)
job_name = "test-job"
result = CliRunner(env={
"LC_ALL": "en_US.UTF-8",
"LANG": "en_US.UTF-8"
}).invoke(
mfscli.commands,
[
"deploy-transform-job",
"--job-name",
job_name,
"--model-uri",
model_s3_uri,
"--input-data-type",
"Some Data Type",
"--input-uri",
"Some Input Uri",
"--content-type",
"Some Content Type",
"--output-path",
"Some Output Path",
"--archive",
],
)
assert result.exit_code == 0
region_name = sagemaker_client.meta.region_name
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
transform_job_description = sagemaker_client.describe_transform_job(
TransformJobName=job_name)
model_name = transform_job_description["ModelName"]
assert model_name in [
model["ModelName"]
for model in sagemaker_client.list_models()["Models"]
]
object_names = [
entry["Key"]
for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert job_name in [
transform_job["TransformJobName"] for transform_job in
sagemaker_client.list_transform_jobs()["TransformJobSummaries"]
]
def test_deploy_cli_creates_sagemaker_and_s3_resources_with_expected_names_and_env_from_s3(
pretrained_model, sagemaker_client):
local_model_path = _download_artifact_from_uri(pretrained_model.model_uri)
artifact_path = "model"
region_name = sagemaker_client.meta.region_name
default_bucket = mfs._get_default_s3_bucket(region_name)
s3_artifact_repo = S3ArtifactRepository("s3://{}".format(default_bucket))
s3_artifact_repo.log_artifacts(local_model_path,
artifact_path=artifact_path)
model_s3_uri = "s3://{bucket_name}/{artifact_path}".format(
bucket_name=default_bucket, artifact_path=pretrained_model.model_path)
app_name = "test-app"
result = CliRunner(env={
"LC_ALL": "en_US.UTF-8",
"LANG": "en_US.UTF-8"
}).invoke(
mfscli.commands,
[
"deploy", "-a", app_name, "-m", model_s3_uri, "--mode",
mfs.DEPLOYMENT_MODE_CREATE
],
)
assert result.exit_code == 0
region_name = sagemaker_client.meta.region_name
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
endpoint_description = sagemaker_client.describe_endpoint(
EndpointName=app_name)
endpoint_production_variants = endpoint_description["ProductionVariants"]
assert len(endpoint_production_variants) == 1
model_name = endpoint_production_variants[0]["VariantName"]
assert model_name in [
model["ModelName"]
for model in sagemaker_client.list_models()["Models"]
]
object_names = [
entry["Key"]
for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert any([
app_name in config["EndpointConfigName"] for config in
sagemaker_client.list_endpoint_configs()["EndpointConfigs"]
])
assert app_name in [
endpoint["EndpointName"]
for endpoint in sagemaker_client.list_endpoints()["Endpoints"]
]
model_environment = sagemaker_client.describe_model(
ModelName=model_name)["PrimaryContainer"]["Environment"]
assert model_environment == {
"MLFLOW_DEPLOYMENT_FLAVOR_NAME": "python_function",
"SERVING_ENVIRONMENT": "SageMaker",
}
def test_deploy_cli_creates_sagemaker_and_s3_resources_with_expected_names_from_s3(
pretrained_model, sagemaker_client):
local_model_path = _download_artifact_from_uri(pretrained_model.model_uri)
artifact_path = "model"
region_name = sagemaker_client.meta.region_name
default_bucket = mfs._get_default_s3_bucket(region_name)
s3_artifact_repo = S3ArtifactRepository('s3://{}'.format(default_bucket))
s3_artifact_repo.log_artifacts(local_model_path,
artifact_path=artifact_path)
model_s3_uri = 's3://{bucket_name}/{artifact_path}'.format(
bucket_name=default_bucket, artifact_path=pretrained_model.model_path)
app_name = "test-app"
result = CliRunner(env={
"LC_ALL": "en_US.UTF-8",
"LANG": "en_US.UTF-8"
}).invoke(mfscli.commands, [
'deploy',
'-a',
app_name,
'-m',
model_s3_uri,
'--mode',
mfs.DEPLOYMENT_MODE_CREATE,
])
assert result.exit_code == 0
region_name = sagemaker_client.meta.region_name
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
endpoint_description = sagemaker_client.describe_endpoint(
EndpointName=app_name)
endpoint_production_variants = endpoint_description["ProductionVariants"]
assert len(endpoint_production_variants) == 1
model_name = endpoint_production_variants[0]["VariantName"]
assert model_name in [
model["ModelName"]
for model in sagemaker_client.list_models()["Models"]
]
object_names = [
entry["Key"]
for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert any([
app_name in config["EndpointConfigName"] for config in
sagemaker_client.list_endpoint_configs()["EndpointConfigs"]
])
assert app_name in [
endpoint["EndpointName"]
for endpoint in sagemaker_client.list_endpoints()["Endpoints"]
]
def test_deploy_creates_sagemaker_and_s3_resources_with_expected_names_from_local(
pretrained_model, sagemaker_client):
app_name = "test-app"
mfs.deploy(app_name=app_name,
model_uri=pretrained_model.model_uri,
mode=mfs.DEPLOYMENT_MODE_CREATE)
region_name = sagemaker_client.meta.region_name
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
endpoint_description = sagemaker_client.describe_endpoint(
EndpointName=app_name)
endpoint_production_variants = endpoint_description["ProductionVariants"]
assert len(endpoint_production_variants) == 1
model_name = endpoint_production_variants[0]["VariantName"]
assert model_name in [
model["ModelName"]
for model in sagemaker_client.list_models()["Models"]
]
object_names = [
entry["Key"]
for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert any([
app_name in config["EndpointConfigName"] for config in
sagemaker_client.list_endpoint_configs()["EndpointConfigs"]
])
assert app_name in [
endpoint["EndpointName"]
for endpoint in sagemaker_client.list_endpoints()["Endpoints"]
]
def test_deploy_cli_creates_sagemaker_and_s3_resources_with_expected_names_and_env_from_local(
pretrained_model, sagemaker_client
):
app_name = "test-app"
region_name = sagemaker_client.meta.region_name
create_sagemaker_deployment_through_cli(app_name, pretrained_model.model_uri, region_name)
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
endpoint_description = sagemaker_client.describe_endpoint(EndpointName=app_name)
endpoint_production_variants = endpoint_description["ProductionVariants"]
assert len(endpoint_production_variants) == 1
model_name = endpoint_production_variants[0]["VariantName"]
assert model_name in [model["ModelName"] for model in sagemaker_client.list_models()["Models"]]
object_names = [
entry["Key"] for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert any(
[
app_name in config["EndpointConfigName"]
for config in sagemaker_client.list_endpoint_configs()["EndpointConfigs"]
]
)
assert app_name in [
endpoint["EndpointName"] for endpoint in sagemaker_client.list_endpoints()["Endpoints"]
]
model_environment = sagemaker_client.describe_model(ModelName=model_name)["PrimaryContainer"][
"Environment"
]
assert model_environment == {
"MLFLOW_DEPLOYMENT_FLAVOR_NAME": "python_function",
"SERVING_ENVIRONMENT": "SageMaker",
}
def test_delete_deployment_synchronous_with_archiving_only_deletes_endpoint(
pretrained_model, sagemaker_client, sagemaker_deployment_client
):
name = "test-app"
region_name = sagemaker_client.meta.region_name
sagemaker_deployment_client.create_deployment(
name=name, model_uri=pretrained_model.model_uri, config=dict(region_name=region_name)
)
sagemaker_deployment_client.delete_deployment(
name=name, config=dict(archive=True, synchronous=True, region_name=region_name)
)
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
s3_objects = s3_client.list_objects_v2(Bucket=default_bucket)
endpoints = sagemaker_client.list_endpoints()
endpoint_configs = sagemaker_client.list_endpoint_configs()
models = sagemaker_client.list_models()
assert s3_objects["KeyCount"] > 0
assert len(endpoints["Endpoints"]) == 0
assert len(endpoint_configs["EndpointConfigs"]) > 0
assert len(models["Models"]) > 0
def test_deploy_creates_sagemaker_transform_job_and_s3_resources_with_expected_names_from_local(
pretrained_model, sagemaker_client):
job_name = "test-job"
mfs.deploy_transform_job(
job_name=job_name,
model_uri=pretrained_model.model_uri,
s3_input_data_type="Some Data Type",
s3_input_uri="Some Input Uri",
content_type="Some Content Type",
s3_output_path="Some Output Path",
archive=True,
)
region_name = sagemaker_client.meta.region_name
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
transform_job_description = sagemaker_client.describe_transform_job(
TransformJobName=job_name)
model_name = transform_job_description["ModelName"]
assert model_name in [
model["ModelName"]
for model in sagemaker_client.list_models()["Models"]
]
object_names = [
entry["Key"]
for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert job_name in [
transform_job["TransformJobName"] for transform_job in
sagemaker_client.list_transform_jobs()["TransformJobSummaries"]
]
def test_deploy_creates_sagemaker_and_s3_resources_with_expected_names(
pretrained_model, sagemaker_client):
app_name = "test-app"
mfs.deploy(app_name=app_name,
model_path=pretrained_model.model_path,
run_id=pretrained_model.run_id,
mode=mfs.DEPLOYMENT_MODE_CREATE)
region_name = sagemaker_client.meta.region_name
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
object_names = [
entry["Key"]
for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([
pretrained_model.run_id in object_name for object_name in object_names
])
assert any([
app_name in model["ModelName"]
for model in sagemaker_client.list_models()["Models"]
])
assert any([
app_name in config["EndpointConfigName"] for config in
sagemaker_client.list_endpoint_configs()["EndpointConfigs"]
])
assert app_name in [
endpoint["EndpointName"]
for endpoint in sagemaker_client.list_endpoints()["Endpoints"]
]
def test_deploy_cli_updates_sagemaker_and_s3_resources_in_replace_mode(
pretrained_model, sagemaker_client
):
app_name = "test-app"
region_name = sagemaker_client.meta.region_name
create_sagemaker_deployment_through_cli(app_name, pretrained_model.model_uri, region_name)
result = CliRunner(env={"LC_ALL": "en_US.UTF-8", "LANG": "en_US.UTF-8"}).invoke(
cli_commands,
[
"update",
"--target",
f"sagemaker:/{region_name}",
"--name",
app_name,
"--model-uri",
pretrained_model.model_uri,
],
)
assert result.exit_code == 0
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
endpoint_description = sagemaker_client.describe_endpoint(EndpointName=app_name)
endpoint_production_variants = endpoint_description["ProductionVariants"]
assert len(endpoint_production_variants) == 1
model_name = endpoint_production_variants[0]["VariantName"]
assert model_name in [model["ModelName"] for model in sagemaker_client.list_models()["Models"]]
object_names = [
entry["Key"] for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert any(
[
app_name in config["EndpointConfigName"]
for config in sagemaker_client.list_endpoint_configs()["EndpointConfigs"]
]
)
assert app_name in [
endpoint["EndpointName"] for endpoint in sagemaker_client.list_endpoints()["Endpoints"]
]
model_environment = sagemaker_client.describe_model(ModelName=model_name)["PrimaryContainer"][
"Environment"
]
assert model_environment == {
"MLFLOW_DEPLOYMENT_FLAVOR_NAME": "python_function",
"SERVING_ENVIRONMENT": "SageMaker",
}
def test_deploy_in_replace_mode_with_archiving_does_not_delete_resources(
pretrained_model, sagemaker_client):
region_name = sagemaker_client.meta.region_name
sagemaker_backend = get_sagemaker_backend(region_name)
sagemaker_backend.set_endpoint_update_latency(5)
app_name = "test-app"
mfs.deploy(app_name=app_name,
model_path=pretrained_model.model_path,
run_id=pretrained_model.run_id,
mode=mfs.DEPLOYMENT_MODE_CREATE)
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
object_names_before_replacement = [
entry["Key"] for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]]
endpoint_configs_before_replacement = [
config["EndpointConfigName"] for config in
sagemaker_client.list_endpoint_configs()["EndpointConfigs"]]
models_before_replacement = [
model["ModelName"] for model in sagemaker_client.list_models()["Models"]]
sk_model = mlflow.sklearn.load_model(
path=pretrained_model.model_path, run_id=pretrained_model.run_id)
new_artifact_path = "model"
with mlflow.start_run():
mlflow.sklearn.log_model(sk_model=sk_model, artifact_path=new_artifact_path)
new_run_id = mlflow.active_run().info.run_uuid
mfs.deploy(app_name=app_name,
model_path=new_artifact_path,
run_id=new_run_id,
mode=mfs.DEPLOYMENT_MODE_REPLACE,
archive=True,
synchronous=True)
object_names_after_replacement = [
entry["Key"] for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]]
endpoint_configs_after_replacement = [
config["EndpointConfigName"] for config in
sagemaker_client.list_endpoint_configs()["EndpointConfigs"]]
models_after_replacement = [
model["ModelName"] for model in sagemaker_client.list_models()["Models"]]
assert all([object_name in object_names_after_replacement
for object_name in object_names_before_replacement])
assert all([endpoint_config in endpoint_configs_after_replacement
for endpoint_config in endpoint_configs_before_replacement])
assert all([model in models_after_replacement for model in models_before_replacement])
def test_create_deployment_creates_sagemaker_and_s3_resources_with_expected_names_and_env_from_s3(
pretrained_model, sagemaker_client, sagemaker_deployment_client
):
local_model_path = _download_artifact_from_uri(pretrained_model.model_uri)
artifact_path = "model"
region_name = sagemaker_client.meta.region_name
default_bucket = mfs._get_default_s3_bucket(region_name)
s3_artifact_repo = S3ArtifactRepository("s3://{}".format(default_bucket))
s3_artifact_repo.log_artifacts(local_model_path, artifact_path=artifact_path)
model_s3_uri = "s3://{bucket_name}/{artifact_path}".format(
bucket_name=default_bucket, artifact_path=pretrained_model.model_path
)
name = "test-app"
sagemaker_deployment_client.create_deployment(
name=name,
model_uri=model_s3_uri,
)
endpoint_description = sagemaker_client.describe_endpoint(EndpointName=name)
endpoint_production_variants = endpoint_description["ProductionVariants"]
assert len(endpoint_production_variants) == 1
model_name = endpoint_production_variants[0]["VariantName"]
assert model_name in [model["ModelName"] for model in sagemaker_client.list_models()["Models"]]
s3_client = boto3.client("s3", region_name=region_name)
object_names = [
entry["Key"] for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert any(
[
name in config["EndpointConfigName"]
for config in sagemaker_client.list_endpoint_configs()["EndpointConfigs"]
]
)
assert name in [
endpoint["EndpointName"] for endpoint in sagemaker_client.list_endpoints()["Endpoints"]
]
model_environment = sagemaker_client.describe_model(ModelName=model_name)["PrimaryContainer"][
"Environment"
]
assert model_environment == {
"MLFLOW_DEPLOYMENT_FLAVOR_NAME": "python_function",
"SERVING_ENVIRONMENT": "SageMaker",
}
def test_deploy_cli_creates_sagemaker_and_s3_resources_with_expected_names_from_local(
pretrained_model, sagemaker_client):
app_name = "test-app"
result = CliRunner(env={
"LC_ALL": "en_US.UTF-8",
"LANG": "en_US.UTF-8"
}).invoke(
mfscli.commands,
[
"deploy",
"-a",
app_name,
"-m",
pretrained_model.model_uri,
"--mode",
mfs.DEPLOYMENT_MODE_CREATE,
],
)
assert result.exit_code == 0
region_name = sagemaker_client.meta.region_name
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
endpoint_description = sagemaker_client.describe_endpoint(
EndpointName=app_name)
endpoint_production_variants = endpoint_description["ProductionVariants"]
assert len(endpoint_production_variants) == 1
model_name = endpoint_production_variants[0]["VariantName"]
assert model_name in [
model["ModelName"]
for model in sagemaker_client.list_models()["Models"]
]
object_names = [
entry["Key"]
for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert any([
app_name in config["EndpointConfigName"] for config in
sagemaker_client.list_endpoint_configs()["EndpointConfigs"]
])
assert app_name in [
endpoint["EndpointName"]
for endpoint in sagemaker_client.list_endpoints()["Endpoints"]
]
def test_deploy_creates_sagemaker_and_s3_resources_with_expected_names_from_s3(
pretrained_model, sagemaker_client):
local_model_path = _download_artifact_from_uri(pretrained_model.model_uri)
artifact_path = "model"
region_name = sagemaker_client.meta.region_name
default_bucket = mfs._get_default_s3_bucket(region_name)
s3_artifact_repo = S3ArtifactRepository("s3://{}".format(default_bucket))
s3_artifact_repo.log_artifacts(local_model_path,
artifact_path=artifact_path)
model_s3_uri = "s3://{bucket_name}/{artifact_path}".format(
bucket_name=default_bucket, artifact_path=pretrained_model.model_path)
app_name = "test-app"
mfs.deploy(app_name=app_name,
model_uri=model_s3_uri,
mode=mfs.DEPLOYMENT_MODE_CREATE)
endpoint_description = sagemaker_client.describe_endpoint(
EndpointName=app_name)
endpoint_production_variants = endpoint_description["ProductionVariants"]
assert len(endpoint_production_variants) == 1
model_name = endpoint_production_variants[0]["VariantName"]
assert model_name in [
model["ModelName"]
for model in sagemaker_client.list_models()["Models"]
]
s3_client = boto3.client("s3", region_name=region_name)
object_names = [
entry["Key"]
for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert any([
app_name in config["EndpointConfigName"] for config in
sagemaker_client.list_endpoint_configs()["EndpointConfigs"]
])
assert app_name in [
endpoint["EndpointName"]
for endpoint in sagemaker_client.list_endpoints()["Endpoints"]
]
def test_deploy_creates_sagemaker_transform_job_and_s3_resources_with_expected_names_from_s3(
pretrained_model, sagemaker_client):
local_model_path = _download_artifact_from_uri(pretrained_model.model_uri)
artifact_path = "model"
region_name = sagemaker_client.meta.region_name
default_bucket = mfs._get_default_s3_bucket(region_name)
s3_artifact_repo = S3ArtifactRepository("s3://{}".format(default_bucket))
s3_artifact_repo.log_artifacts(local_model_path,
artifact_path=artifact_path)
model_s3_uri = "s3://{bucket_name}/{artifact_path}".format(
bucket_name=default_bucket, artifact_path=pretrained_model.model_path)
job_name = "test-job"
mfs.deploy_transform_job(
job_name=job_name,
model_uri=model_s3_uri,
s3_input_data_type="Some Data Type",
s3_input_uri="Some Input Uri",
content_type="Some Content Type",
s3_output_path="Some Output Path",
archive=True,
)
transform_job_description = sagemaker_client.describe_transform_job(
TransformJobName=job_name)
model_name = transform_job_description["ModelName"]
assert model_name in [
model["ModelName"]
for model in sagemaker_client.list_models()["Models"]
]
s3_client = boto3.client("s3", region_name=region_name)
object_names = [
entry["Key"]
for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any([model_name in object_name for object_name in object_names])
assert job_name in [
transform_job["TransformJobName"] for transform_job in
sagemaker_client.list_transform_jobs()["TransformJobSummaries"]
]
def test_deploy_cli_creates_sagemaker_transform_job_and_s3_resources_with_expected_names_from_local(
pretrained_model, sagemaker_client
):
job_name = "test-job"
result = CliRunner(env={"LC_ALL": "en_US.UTF-8", "LANG": "en_US.UTF-8"}).invoke(
mfscli.commands,
[
"deploy-transform-job",
"--job-name",
job_name,
"--model-uri",
pretrained_model.model_uri,
"--input-data-type",
"Some Data Type",
"--input-uri",
"Some Input Uri",
"--content-type",
"Some Content Type",
"--output-path",
"Some Output Path",
"--archive",
],
)
assert result.exit_code == 0
region_name = sagemaker_client.meta.region_name
s3_client = boto3.client("s3", region_name=region_name)
default_bucket = mfs._get_default_s3_bucket(region_name)
transform_job_description = sagemaker_client.describe_transform_job(TransformJobName=job_name)
model_name = transform_job_description["ModelName"]
assert model_name in [model["ModelName"] for model in sagemaker_client.list_models()["Models"]]
object_names = [
entry["Key"] for entry in s3_client.list_objects(Bucket=default_bucket)["Contents"]
]
assert any(model_name in object_name for object_name in object_names)
assert job_name in [
transform_job["TransformJobName"]
for transform_job in sagemaker_client.list_transform_jobs()["TransformJobSummaries"]
]
