address pr comments

Author: jesterhazy

.gitignore

@@ -25,4 +25,4 @@ venv/
 *~
 .pytest_cache/
 *.swp
-tests/data/local_mode_lock
+.docker/

tests/data/local_mode_lock

@@ -0,0 +1,5 @@
+
+recursive-include . *
+recursive-exclude . __pycache__*
+recursive-exclude . *.pyc
+recursive-exclude . *.pyo

tests/data/mxnet_mnist/MANIFEST.in

@@ -0,0 +1,3 @@
+
+[wheel]
+universal = 1

tests/data/mxnet_mnist/setup.cfg

@@ -0,0 +1,6 @@
+
+from setuptools import setup
+setup(packages=[''],
+      name="mnist",
+      version='1.0.0',
+      include_package_data=True)

tests/data/mxnet_mnist/setup.py

@@ -23,6 +23,10 @@
 
 @contextmanager
 def lock(path=DEFAULT_LOCK_PATH):
+    """Create a file lock to control concurrent test execution. Certain tests or
+    test operations need to limit concurrency to work reliably. Examples include
+    local mode endpoint tests and vpc creation tests.
+    """
     f = open(path, 'w')
     fd = f.fileno()
 

tests/integ/lock.py

@@ -28,6 +28,7 @@
 from sagemaker.mxnet import MXNet
 from sagemaker.tensorflow import TensorFlow
 
+# endpoint tests all use the same port, so we use this lock to prevent concurrent execution
 LOCK_PATH = os.path.join(tempfile.gettempdir(), 'sagemaker_test_local_mode_lock')
 DATA_PATH = os.path.join(DATA_DIR, 'iris', 'data')
 DEFAULT_REGION = 'us-west-2'

tests/integ/test_local_mode.py

@@ -37,6 +37,7 @@ def test_source_dirs(tmpdir, sagemaker_local_session):
                         sagemaker_session=sagemaker_local_session)
     estimator.fit()
 
+    # endpoint tests all use the same port, so we use this lock to prevent concurrent execution
     with lock.lock():
         try:
             predictor = estimator.deploy(initial_instance_count=1, instance_type='local')

tests/integ/test_source_dirs.py

@@ -93,6 +93,7 @@ def _create_vpc_with_name(ec2_client, region, name):
 
 
 def get_or_create_vpc_resources(ec2_client, region, name=VPC_NAME):
+    # use lock to prevent race condition when tests are running concurrently
     with lock.lock(LOCK_PATH):
         if _vpc_exists(ec2_client, name):
             print('using existing vpc: {}'.format(name))