Correctly handle TooManyBuckets error_code in default_bucket method

[jalabort]

At my organisation we have reached our maximum number of allowed buckets on s3 and as a direct consequence of that we started getting the following error when using the sagemaker-python-sdk:

INFO:botocore.vendored.requests.packages.urllib3.connectionpool:Starting new HTTPS connection (1): s3.amazonaws.com
Traceback (most recent call last):
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/bin/sagemaker", line 6, in <module>
    exec(compile(open(__file__).read(), __file__, 'exec'))
  File "/Users/joan.alabort/cvdev/beatrix/hudl_beatrix/cli/bin/sagemaker", line 6, in <module>
    fire.Fire(component=SageMakerRunner)
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/fire/core.py", line 127, in Fire
    component_trace = _Fire(component, args, context, name)
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/fire/core.py", line 366, in _Fire
    component, remaining_args)
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/fire/core.py", line 542, in _CallCallable
    result = fn(*varargs, **kwargs)
  File "/Users/joan.alabort/cvdev/beatrix/hudl_beatrix/sagemaker/base.py", line 314, in train
    self._train(source_local_path, dataset_s3_path, job_name)
  File "/Users/joan.alabort/cvdev/beatrix/hudl_beatrix/sagemaker/base.py", line 364, in _train
    run_tensorboard_locally=self._run_tensorboard_locally
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/sagemaker/tensorflow/estimator.py", line 162, in fit
    fit_super()
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/sagemaker/tensorflow/estimator.py", line 154, in fit_super
    super(TensorFlow, self).fit(inputs, wait, logs, job_name)
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/sagemaker/estimator.py", line 498, in fit
    code_bucket = self.sagemaker_session.default_bucket()
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/sagemaker/session.py", line 159, in default_bucket
    s3.create_bucket(Bucket=default_bucket)
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/boto3/resources/factory.py", line 520, in do_action
    response = action(self, *args, **kwargs)
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/boto3/resources/action.py", line 83, in __call__
    response = getattr(parent.meta.client, operation_name)(**params)
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/botocore/client.py", line 317, in _api_call
    return self._make_api_call(operation_name, kwargs)
  File "/Users/joan.alabort/miniconda3/envs/beatrix_27/lib/python2.7/site-packages/botocore/client.py", line 615, in _make_api_call
    raise error_class(parsed_response, operation_name)
botocore.exceptions.ClientError: An error occurred (TooManyBuckets) when calling the CreateBucket operation: You have attempted to create more buckets than allowed

This little patch seems to take care of this issue.

I could not find much information in terms of the expected PR structure so I hope this suffices, happy to make the pertinent changes if it does not.

[owen-t]

Thanks for your submission!

This looks good to me.

[owen-t]

I don't think we need this logging message, in the normal case we don't log this.

[owen-t]

This is a little esoteric - if we remove the logging below, can you add some developer documentation:

# Succeed if the default bucket exists

[jalabort]

@owen-t Addressed your comments above.

[jalabort]

Is there something I need to amend due to the failing teamcity build? I cannot see what's wrong with it.

[ChoiByungWook]

Hello,
We don't publish our build logs externally. A test failed, but the issue is resolved and we're rerunning your build. We will merge it in once the build is complete. Thanks again for your contribution!