AWS SageMaker SparkML架构：成员.环境“”未能满足约束

Question

我正在通过Sagemaker将模型部署到AWS上：

我将JSON模式设置如下：

import json
schema = {
    "input": [
        {
            "name": "V1",
            "type": "double"
        }, 
        {
            "name": "V2",
            "type": "double"
        }, 
        {
            "name": "V3",
            "type": "double"
        }, 
        {
            "name": "V4",
            "type": "double"
        }, 
        {
            "name": "V5",
            "type": "double"
        }, 
        {
            "name": "V6",
            "type": "double"
        },
        {
            "name": "V7",
            "type": "double"
        }, 
        {
            "name": "V8",
            "type": "double"
        }, 
        {
            "name": "V9",
            "type": "double"
        }, 
        {
            "name": "V10",
            "type": "double"
        }, 
        {
            "name": "V11",
            "type": "double"
        }, 
        {
            "name": "V12",
            "type": "double"
        }, 
        {
            "name": "V13",
            "type": "double"
        }, 
        {
            "name": "V14",
            "type": "double"
        },
        {
            "name": "V15",
            "type": "double"
        }, 
        {
            "name": "V16",
            "type": "double"
        }, 
        {
            "name": "V17",
            "type": "double"
        }, 
        {
            "name": "V18",
            "type": "double"
        }, 
        {
            "name": "V19",
            "type": "double"
        }, 
                {
            "name": "V20",
            "type": "double"
        }, 
        {
            "name": "V21",
            "type": "double"
        }, 
        {
            "name": "V22",
            "type": "double"
        },
        {
            "name": "V23",
            "type": "double"
        }, 
        {
            "name": "V24",
            "type": "double"
        }, 
        {
            "name": "V25",
            "type": "double"
        }, 
        {
            "name": "V26",
            "type": "double"
        }, 
        {
            "name": "V27",
            "type": "double"
        },
        {
            "name": "V28",
            "type": "double"
        },
        {
            "name": "Amount",
            "type": "double"
        },         
    ],
    "output": 
        {
            "name": "features",
            "type": "double",
            "struct": "vector"
        }
}
schema_json = json.dumps(schema)
print(schema_json)

并部署为：

from sagemaker.model import Model
from sagemaker.pipeline import PipelineModel
from sagemaker.sparkml.model import SparkMLModel

sparkml_data = 's3://{}/{}/{}'.format(s3_model_bucket, s3_model_key_prefix, 'model.tar.gz')
# passing the schema defined above by using an environment variable that sagemaker-sparkml-serving understands
sparkml_model = SparkMLModel(model_data=sparkml_data, env={'SAGEMAKER_SPARKML_SCHEMA' : schema_json})
xgb_model = Model(model_data=xgb_model.model_data, image=training_image)

model_name = 'inference-pipeline-' + timestamp_prefix
sm_model = PipelineModel(name=model_name, role=role, models=[sparkml_model, xgb_model])

    endpoint_name = 'inference-pipeline-ep-' + timestamp_prefix
sm_model.deploy(initial_instance_count=1, instance_type='ml.c4.xlarge', endpoint_name=endpoint_name)

我得到的错误如下：

ClientError:调用CreateModel操作时发生错误（ValidationException）：检测到1个验证错误：值“{SAGEMAKER_SPARKML_SCHEMA={”输入“：[{”类型“：“double”，“名称“：“V1”}，{”类型“：“V2”}，{”类型“：“V3”}，{”类型“：“double”，“名称“：“V4”}，{”类型“：“double”，“名称“：“V5”}，{”类型“：”“double”，“name”：“V6”}，{“type”：“double”，“name”：“V7”}，{“type”：“double”，“name”：“V9”}，{“type”：“double”，“name”：“V10”}，{“type”：“double”，“name”：“V11”}，{“type”：“double”，“name”：“V12”}，{“type”：“double”，“name”：“V13”}，{“type”：“double”，“name”：“V14”}，{“type”：“double”，“name”：“name”：“V15”}，{“type”：“type”：双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度，双精度double、name:“V26”}、{“type”：“double”、“name”：“V27”}、{“type”：“double”、“name”：“V28”}、{“type”：“double”、“name”：“Amount”}]、“output”：{“type”：“double”、“name”：“features”、“struct”：“vector”}}at“containers.1**.member.environment”未能满足约束：映射值必须满足约束：[成员的长度必须小于或等于1024，**成员的长度必须大于或等于0，成员必须满足正则表达式模式：[\S\S]*]

我尝试将我的功能减少到20个，并且它能够部署。只是想知道如何传递具有29个属性的模式？

Answer 1

我不认为1024限制的环境长度会在短时间内增加。为了解决这个问题，您可以尝试使用SAGEMAKER_SPARKML_SCHEMAenv var重建Spark ml容器：

https://github.com/aws/sagemaker-sparkml-serving-container/blob/master/README.md#running-the-image-locally