Skip to content

Fix: Add additional model builder telemetry #4334

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 5 commits into from
Dec 20, 2023
Merged
Show file tree
Hide file tree
Changes from 3 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
35 changes: 30 additions & 5 deletions src/sagemaker/serve/utils/telemetry_logger.py
Original file line number Diff line number Diff line change
Expand Up @@ -59,21 +59,35 @@ def wrapper(self, *args, **kwargs):
caught_ex = None

image_uri_tail = self.image_uri.split("/")[1]
extra = f"{func_name}&{MODEL_SERVER_TO_CODE[str(self.model_server)]}&{image_uri_tail}"
extra = (
f"{func_name}"
f"&x-modelServer={MODEL_SERVER_TO_CODE[str(self.model_server)]}"
f"&x-imageTag={image_uri_tail}"
)

if self.model_server == ModelServer.DJL_SERVING or self.model_server == ModelServer.TGI:
extra += f"&{self.model}"
extra += f"&x-modelName={self.model}"

try:
response = func(self, *args, **kwargs)
if not self.serve_settings.telemetry_opt_out:
_send_telemetry(
"1", MODE_TO_CODE[str(self.mode)], self.sagemaker_session, None, extra
"1",
MODE_TO_CODE[str(self.mode)],
self.sagemaker_session,
None,
None,
extra,
)
except ModelBuilderException as e:
if not self.serve_settings.telemetry_opt_out:
_send_telemetry(
"0", MODE_TO_CODE[str(self.mode)], self.sagemaker_session, str(e), extra
"0",
MODE_TO_CODE[str(self.mode)],
self.sagemaker_session,
str(e),
e.__class__.__name__,
extra,
)
caught_ex = e
except Exception as e: # pylint: disable=W0703
Expand All @@ -93,13 +107,22 @@ def _send_telemetry(
mode: int,
session: Session,
failure_reason: str = None,
failure_type: str = None,
extra_info: str = None,
) -> None:
"""Make GET request to an empty object in S3 bucket"""
try:
accountId = _get_accountId(session)
region = _get_region_or_default(session)
url = _construct_url(accountId, str(mode), status, failure_reason, extra_info, region)
url = _construct_url(
accountId,
str(mode),
status,
failure_reason,
failure_type,
extra_info,
region,
)
_requests_helper(url, 2)
logger.debug("ModelBuilder metrics emitted.")
except Exception: # pylint: disable=W0703
Expand All @@ -111,6 +134,7 @@ def _construct_url(
mode: str,
status: str,
failure_reason: str,
failure_type: str,
extra_info: str,
region: str,
) -> str:
Expand All @@ -124,6 +148,7 @@ def _construct_url(
)
if failure_reason:
base_url += f"&x-failureReason={failure_reason}"
base_url += f"&x-failureType={failure_type}"
if extra_info:
base_url += f"&x-extra={extra_info}"
return base_url
Expand Down
154 changes: 146 additions & 8 deletions tests/unit/sagemaker/serve/utils/test_telemetry_logger.py
Original file line number Diff line number Diff line change
Expand Up @@ -12,19 +12,46 @@
# language governing permissions and limitations under the License.
from __future__ import absolute_import
import unittest
from unittest.mock import MagicMock, patch
from sagemaker.serve.utils.telemetry_logger import _send_telemetry
from unittest.mock import Mock, patch
from sagemaker.serve import Mode, ModelServer
from sagemaker.serve.utils.telemetry_logger import (
_send_telemetry,
_capture_telemetry,
_construct_url,
)
from sagemaker.serve.utils.exceptions import ModelBuilderException, LocalModelOutOfMemoryException

mock_session = MagicMock()
MOCK_SESSION = Mock()
MOCK_FUNC_NAME = "Mock.deploy"
MOCK_DJL_CONTAINER = (
"763104351884.dkr.ecr.us-west-2.amazonaws.com/" "djl-inference:0.25.0-deepspeed0.11.0-cu118"
)
MOCK_TGI_CONTAINER = (
"763104351884.dkr.ecr.us-east-1.amazonaws.com/"
"huggingface-pytorch-inference:2.0.0-transformers4.28.1-cpu-py310-ubuntu20.04"
)
MOCK_HUGGINGFACE_ID = "meta-llama/Llama-2-7b-hf"
MOCK_EXCEPTION = LocalModelOutOfMemoryException("mock raise ex")


class ModelBuilderMock:
def __init__(self):
self.serve_settings = Mock()
self.sagemaker_session = MOCK_SESSION

@_capture_telemetry(MOCK_FUNC_NAME)
def mock_deploy(self, mock_exception_func=None):
if mock_exception_func:
mock_exception_func()


class TestTelemetryLogger(unittest.TestCase):
@patch("sagemaker.serve.utils.telemetry_logger._requests_helper")
@patch("sagemaker.serve.utils.telemetry_logger._get_accountId")
def test_log_sucessfully(self, mocked_get_accountId, mocked_request_helper):
mock_session.boto_session.region_name = "ap-south-1"
MOCK_SESSION.boto_session.region_name = "ap-south-1"
mocked_get_accountId.return_value = "testAccountId"
_send_telemetry("someStatus", 1, mock_session)
_send_telemetry("someStatus", 1, MOCK_SESSION)
mocked_request_helper.assert_called_with(
"https://dev-exp-t-ap-south-1.s3.ap-south-1.amazonaws.com/"
"telemetry?x-accountId=testAccountId&x-mode=1&x-status=someStatus",
Expand All @@ -34,9 +61,120 @@ def test_log_sucessfully(self, mocked_get_accountId, mocked_request_helper):
@patch("sagemaker.serve.utils.telemetry_logger._get_accountId")
def test_log_handle_exception(self, mocked_get_accountId):
mocked_get_accountId.side_effect = Exception("Internal error")
_send_telemetry("someStatus", 1, mock_session)
_send_telemetry("someStatus", 1, MOCK_SESSION)
self.assertRaises(Exception)

@patch("sagemaker.serve.utils.telemetry_logger._send_telemetry")
def test_capture_telemetry_decorator_djl_success(self, mock_send_telemetry):
mock_model_builder = ModelBuilderMock()
mock_model_builder.serve_settings.telemetry_opt_out = False
mock_model_builder.image_uri = MOCK_DJL_CONTAINER
mock_model_builder.model = MOCK_HUGGINGFACE_ID
mock_model_builder.mode = Mode.LOCAL_CONTAINER
mock_model_builder.model_server = ModelServer.DJL_SERVING

mock_model_builder.mock_deploy()

expected_extra_str = (
f"{MOCK_FUNC_NAME}"
"&x-modelServer=4"
"&x-imageTag=djl-inference:0.25.0-deepspeed0.11.0-cu118"
f"&x-modelName={MOCK_HUGGINGFACE_ID}"
)
mock_send_telemetry.assert_called_once_with(
"1", 2, MOCK_SESSION, None, None, expected_extra_str
)

@patch("sagemaker.serve.utils.telemetry_logger._send_telemetry")
def test_capture_telemetry_decorator_tgi_success(self, mock_send_telemetry):
mock_model_builder = ModelBuilderMock()
mock_model_builder.serve_settings.telemetry_opt_out = False
mock_model_builder.image_uri = MOCK_TGI_CONTAINER
mock_model_builder.model = MOCK_HUGGINGFACE_ID
mock_model_builder.mode = Mode.LOCAL_CONTAINER
mock_model_builder.model_server = ModelServer.TGI

mock_model_builder.mock_deploy()

expected_extra_str = (
f"{MOCK_FUNC_NAME}"
"&x-modelServer=6"
"&x-imageTag=huggingface-pytorch-inference:2.0.0-transformers4.28.1-cpu-py310-ubuntu20.04"
f"&x-modelName={MOCK_HUGGINGFACE_ID}"
)
mock_send_telemetry.assert_called_once_with(
"1", 2, MOCK_SESSION, None, None, expected_extra_str
)

@patch("sagemaker.serve.utils.telemetry_logger._send_telemetry")
def test_capture_telemetry_decorator_no_call_when_disabled(self, mock_send_telemetry):
mock_model_builder = ModelBuilderMock()
mock_model_builder.serve_settings.telemetry_opt_out = True
mock_model_builder.image_uri = MOCK_DJL_CONTAINER
mock_model_builder.model = MOCK_HUGGINGFACE_ID
mock_model_builder.model_server = ModelServer.DJL_SERVING

if __name__ == "__main__":
unittest.main()
mock_model_builder.mock_deploy()

assert not mock_send_telemetry.called

@patch("sagemaker.serve.utils.telemetry_logger._send_telemetry")
def test_capture_telemetry_decorator_handle_exception_success(self, mock_send_telemetry):
mock_model_builder = ModelBuilderMock()
mock_model_builder.serve_settings.telemetry_opt_out = False
mock_model_builder.image_uri = MOCK_DJL_CONTAINER
mock_model_builder.model = MOCK_HUGGINGFACE_ID
mock_model_builder.mode = Mode.LOCAL_CONTAINER
mock_model_builder.model_server = ModelServer.DJL_SERVING

mock_exception = Mock()
mock_exception_obj = MOCK_EXCEPTION
mock_exception.side_effect = mock_exception_obj

with self.assertRaises(ModelBuilderException) as _:
mock_model_builder.mock_deploy(mock_exception)

expected_extra_str = (
f"{MOCK_FUNC_NAME}"
"&x-modelServer=4"
"&x-imageTag=djl-inference:0.25.0-deepspeed0.11.0-cu118"
f"&x-modelName={MOCK_HUGGINGFACE_ID}"
)
mock_send_telemetry.assert_called_once_with(
"0",
2,
MOCK_SESSION,
str(mock_exception_obj),
mock_exception_obj.__class__.__name__,
expected_extra_str,
)

def test_construct_url_with_failure_reason_and_extra_info(self):
mock_accountId = "12345678910"
mock_mode = Mode.LOCAL_CONTAINER
mock_status = "0"
mock_failure_reason = str(MOCK_EXCEPTION)
mock_failure_type = MOCK_EXCEPTION.__class__.__name__
mock_extra_info = "mock_extra_info"
mock_region = "us-west-2"

ret_url = _construct_url(
accountId=mock_accountId,
mode=mock_mode,
status=mock_status,
failure_reason=mock_failure_reason,
failure_type=mock_failure_type,
extra_info=mock_extra_info,
region=mock_region,
)

expected_base_url = (
f"https://dev-exp-t-{mock_region}.s3.{mock_region}.amazonaws.com/telemetry?"
f"x-accountId={mock_accountId}"
f"&x-mode={mock_mode}"
f"&x-status={mock_status}"
f"&x-failureReason={mock_failure_reason}"
f"&x-failureType={mock_failure_type}"
f"&x-extra={mock_extra_info}"
)
self.assertEquals(ret_url, expected_base_url)