在本地

时间:2017-02-28 00:00:23

标签: google-cloud-ml google-cloud-ml-engine

我已按照以下说明在LOCAL上设置google-cloud-ml:MAC / LINUX

google-cloud-ml setup

但是在使用此命令验证设置时,我收到以下错误

curl https://raw.githubusercontent.com/GoogleCloudPlatform/cloudml-samples/master/tools/check_environment.py | python
  

追踪(最近一次通话):     文件“”,第70行,in     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/init.py”,第16行,in       来自google.cloud.ml.dataflow._analyzer import AnalyzeModel     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/dataflow/init.py”,第22行,in       来自_ml_transforms导入DeployVersion     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/dataflow/_ml_transforms.py”,第24行,       将_ml_functions导入为ml_func     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/dataflow/_ml_functions.py”,第25行,在       来自google.cloud.ml.io.coders导入TrainingJobResult     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/io/init.py”,第21行,在       来自转换导入LoadFeatures     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/io/transforms.py”,第23行,in       来自google.cloud.ml.dataflow.io import tfrecordio     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/dataflow/io/init.py”,第15行,在       import tfrecordio     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/dataflow/io/tfrecordio.py”,第16行,in       _crc32c_fn = snappy._crc32c #pylint:disable = protected-access   AttributeError:'module'对象没有属性'_crc32c'

看起来谷歌云的数据流库存在一些问题。

我尝试使用以下命令升级数据流库

pip install --upgrade google-api-python-client

但现在,另一个错误。以下是堆栈跟踪:

  

追踪(最近一次通话):     文件“”,第70行,in     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/init.py”,第16行,in       来自google.cloud.ml.dataflow._analyzer import AnalyzeModel     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/dataflow/init.py”,第17行,in       来自_analyzer import AnalyzeModel     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/google/cloud/ml/dataflow/_analyzer.py”,第19行,in       将apache_beam导入为beam     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apache_beam/init.py”,第78行,in       来自apache_beam import io     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apache_beam/io/init.py”,第21行,in       来自apache_beam.io.avroio import *     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apache_beam/io/avroio.py”,第29行,       从apache_beam.io导入filebasedsource     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apache_beam/io/filebasedsource.py”,第32行,in       来自apache_beam.io import concat_source     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apache_beam/io/concat_source.py”,第24行,在       来自apache_beam.io import iobase     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apache_beam/io/iobase.py”,第853行,in       来自apache_beam.runners.dataflow.native_io.iobase import *     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apache_beam/runners/init.py”,第23行,in       来自apache_beam.runners.dataflow_runner导入DataflowRunner     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apache_beam/runners/dataflow_runner.py”,第32行,in       来自apache_beam.internal import json_value     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apache_beam/internal/json_value.py”,第20行,in       来自apitools.base.py import extra_types     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apitools/base/py/init.py”,第23行,在       来自apitools.base.py.credentials_lib import *     文件“/Users/pratyusha/miniconda2/envs/cloudml/lib/python2.7/site-packages/apitools/base/py/credentials_lib.py”,第50行,in       来自oauth2client import locked_file

1 个答案:

答案 0 :(得分:0)

这看起来像是'snappy'的不同版本。你能检查一下你安装了哪个版本的python-snappy吗?

您可以通过运行:

来执行此操作
pip freeze | grep python-snappy

它适用于我的版本0.5,这是最新版本。如果您有早期版本,请尝试运行:

pip install --upgrade python-snappy

(必要时使用'sudo'或'--user'。)