Calls to `create_dataset` and `get_table` are timing out

[tswast]

We're getting some frequent test failures in the pandas integration tests due to timed out requests, possibly due to googleapis/google-cloud-python#10219 which added non-None default timeout.

It seems 60 seconds is not enough for these calls (but strange that so many API requests are timing out).

Environment details

Context: googleapis/python-bigquery-pandas#309

Steps to reproduce

1. Make frequent calls to create_dataset.
2. Observe flakey tests.

Stack trace

        if new_retry.is_exhausted():
>           raise MaxRetryError(_pool, url, error or ResponseError(cause))
E           urllib3.exceptions.MaxRetryError: HTTPSConnectionPool(host='bigquery.googleapis.com', port=443): Max retries exceeded with url: /bigquery/v2/projects/pandas-travis/datasets/pydata_pandas_bq_testing_dddqwbgrjz/tables/hrsctwrmva (Caused by NewConnectionError('<urllib3.connection.VerifiedHTTPSConnection object at 0x7ff87946e898>: Failed to establish a new connection: [Errno 110] Connection timed out',))
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/urllib3/util/retry.py:436: MaxRetryError
During handling of the above exception, another exception occurred:
self = <pandas.tests.io.test_gbq.TestToGBQIntegrationWithServiceAccountKeyPath object at 0x7ff879725160>
gbq_dataset = 'pydata_pandas_bq_testing_dddqwbgrjz.hrsctwrmva'
    def test_roundtrip(self, gbq_dataset):
        destination_table = gbq_dataset
    
        test_size = 20001
        df = make_mixed_dataframe_v2(test_size)
    
        df.to_gbq(
            destination_table,
            _get_project_id(),
            chunksize=None,
>           credentials=_get_credentials(),
        )
pandas/tests/io/test_gbq.py:185: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
pandas/core/frame.py:1551: in to_gbq
    credentials=credentials,
pandas/io/gbq.py:219: in to_gbq
    private_key=private_key,
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/pandas_gbq/gbq.py:1202: in to_gbq
    if table.exists(table_id):
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/pandas_gbq/gbq.py:1312: in exists
    self.client.get_table(table_ref)
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/google/cloud/bigquery/client.py:679: in get_table
    retry, method="GET", path=table_ref.path, timeout=timeout
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/google/cloud/bigquery/client.py:556: in _call_api
    return call()
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/google/api_core/retry.py:286: in retry_wrapped_func
    on_error=on_error,
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/google/api_core/retry.py:184: in retry_target
    return target()
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/google/cloud/_http.py:419: in api_request
    timeout=timeout,
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/google/cloud/_http.py:277: in _make_request
    method, url, headers, data, target_object, timeout=timeout
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/google/cloud/_http.py:315: in _do_request
    url=url, method=method, headers=headers, data=data, timeout=timeout
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/google/auth/transport/requests.py:317: in request
    **kwargs
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/requests/sessions.py:533: in request
    resp = self.send(prep, **send_kwargs)
../../../miniconda3/envs/pandas-dev/lib/python3.6/site-packages/requests/sessions.py:646: in send
    r = adapter.send(request, **kwargs)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

[shollyman]

Do the tests favor a particular region? Always running in US multiregion?

[tswast]

Just the default US region. None of the to_gbq tests at https://github.com/pandas-dev/pandas/blob/8105a7e8282edba5c98138ac09ed2c3bec7823e6/pandas/tests/io/test_gbq.py explicitly set a region.

[meredithslota]

Assigning Seth but if this is something @plamut could look at, let's discuss.

[plamut]

I'm currently working on #40 which might be similar. Once that is done, we can re-check and see if it also helped with this issue.

[meredithslota]

#40 seems closed now; @plamut can you take a look?