AWS Sagemaker Model 进行预测器调用时出错

Question

I'm trying to make a tensorFlow predictor call but I am getting a ModelError - 502 Bad Gateway.我正在尝试进行 tensorFlow 预测器调用，但我收到 ModelError - 502 Bad Gateway。 I can't seem to trace back to what is going on in the server that is causing this error.我似乎无法追溯到导致此错误的服务器中发生的事情。 This model was deployed on ml.c4.xlarge instances.此 model 部署在 ml.c4.xlarge 实例上。

As requested from comments, here is the code used to deploy the model:根据评论的要求，这里是用于部署 model 的代码：

#Import Tensorflow Model
from sagemaker.tensorflow.serving import Model
sagemaker_model = Model(model_data='s3://' + sagemaker_session.default_bucket() + '/Scikit-keras-NLP-pipeline-examplet/train/example.tar.gz',
                        role=role,
                        sagemaker_session=sagemaker_session)

scikit_learn_inference_model = sklearn_preprocessor.create_model() 

#Build Inference Pipeline
sm_model = PipelineModel(
    name=model_name, 
    role=role, 
    models=[
        scikit_learn_inference_model, 
        sagemaker_model],
    sagemaker_session=sagemaker_session)

sm_model.deploy(initial_instance_count=1, instance_type='ml.c4.xlarge', endpoint_name=endpoint_name)

For reference: here is the code in a Sagemaker notebook that is using a.csv file saved in S3 to try and make the prediction:供参考：这是 Sagemaker 笔记本中的代码，该笔记本使用保存在 S3 中的 .csv 文件尝试进行预测：

from sagemaker.predictor import json_serializer, csv_serializer, json_deserializer, RealTimePredictor
from sagemaker.content_types import CONTENT_TYPE_CSV, CONTENT_TYPE_JSON

predictor = RealTimePredictor(
    endpoint='example',
    sagemaker_session=sagemaker_session,
    serializer=csv_serializer,
    content_type=CONTENT_TYPE_CSV,
    accept=CONTENT_TYPE_JSON)

from sklearn.pipeline import Pipeline
from sklearn.preprocessing import Binarizer, StandardScaler, OneHotEncoder
from sklearn.impute import SimpleImputer

column_names = ['CRIM', 'ZN', 'INDUS', 'CHAS', 'NOX', 'RM', 'AGE', 'DIS', 'RAD', 'TAX', 'PTRATIO', 'B', 'LSTAT', 'MEDV']
dff = pd.read_csv('housing.csv',delimiter=r"\s+", names=column_names)
dff.drop('MEDV',axis=1,inplace=True)

#String
x = '0.00632,18.0,2.31,0,0.538,6.575,65.2,4.09,1,296.0,15.3,396.9,4.98'

#DataFrame
y= dff.head(1)

#Array
z = np.array([[0.00632,18.0,2.31,0,0.538,6.575,65.2,4.09,1,296.0,15.3,396.9,4.98]])

print(predictor.predict(z))

Here is the full error for reference:这是供参考的完整错误：

---------------------------------------------------------------------------
ModelError                                Traceback (most recent call last)
<ipython-input-33-1a95d0a95a02> in <module>()
----> 1 print(predictor.predict(z))

~/anaconda3/envs/tensorflow_p36/lib/python3.6/site-packages/sagemaker/predictor.py in predict(self, data, initial_args, target_model)
    108 
    109         request_args = self._create_request_args(data, initial_args, target_model)
--> 110         response = self.sagemaker_session.sagemaker_runtime_client.invoke_endpoint(**request_args)
    111         return self._handle_response(response)
    112 

~/anaconda3/envs/tensorflow_p36/lib/python3.6/site-packages/botocore/client.py in _api_call(self, *args, **kwargs)
    314                     "%s() only accepts keyword arguments." % py_operation_name)
    315             # The "self" in this scope is referring to the BaseClient.
--> 316             return self._make_api_call(operation_name, kwargs)
    317 
    318         _api_call.__name__ = str(py_operation_name)

~/anaconda3/envs/tensorflow_p36/lib/python3.6/site-packages/botocore/client.py in _make_api_call(self, operation_name, api_params)
    624             error_code = parsed_response.get("Error", {}).get("Code")
    625             error_class = self.exceptions.from_code(error_code)
--> 626             raise error_class(parsed_response, operation_name)
    627         else:
    628             return parsed_response

ModelError: An error occurred (ModelError) when calling the InvokeEndpoint operation: Received server error (502) from container-2 with message "<html>
<head><title>502 Bad Gateway</title></head>
<body>
<center><h1>502 Bad Gateway</h1></center>
<hr><center>nginx/1.16.1</center>
</body>
</html>

Finally, a snippet from the CloudWatch Logs if this helps:最后，来自 CloudWatch Logs 的片段（如果有帮助）：

2020/07/21 06:31:23 [error] 33#33: *14 connect() failed (111: Connection refused) while connecting to upstream, client: 10.32.0.3, server: , request: "POST /invocations HTTP/1.1", subrequest: "/v1/models/Servo:predict", upstream: "http://127.0.0.1:22001/v1/models/Servo:predict", host: "container-2.aws.local:9553"

Answer 1

Invoke/Deploy with serializer.使用序列化程序调用/部署。

from sagemaker.serializers import IdentitySerializer
    
sm_model.deploy(initial_instance_count=1, instance_type='ml.c4.xlarge', endpoint_name=endpoint_name, serializer=IdentitySerializer)

AWS Sagemaker Model 进行预测器调用时出错

问题描述

1 个解决方案

解决方案1
0

AWS Sagemaker Model 进行预测器调用时出错

问题描述

1 个解决方案

解决方案1 0

解决方案1
0