def setUp(self):
self.remote_log_group = 'log_group_name'
self.region_name = 'us-west-2'
self.local_log_location = 'local/log/location'
self.filename_template = '{dag_id}/{task_id}/{execution_date}/{try_number}.log'
self.cloudwatch_task_handler = CloudwatchTaskHandler(
self.local_log_location,
f"arn:aws:logs:{self.region_name}:11111111:log-group:{self.remote_log_group}",
self.filename_template,
)
self.cloudwatch_task_handler.hook
date = datetime(2020, 1, 1)
dag_id = 'dag_for_testing_file_task_handler'
task_id = 'task_for_testing_file_log_handler'
self.dag = DAG(dag_id=dag_id, start_date=date)
task = DummyOperator(task_id=task_id, dag=self.dag)
self.ti = TaskInstance(task=task, execution_date=date)
self.ti.try_number = 1
self.ti.state = State.RUNNING
self.remote_log_stream = '{}/{}/{}/{}.log'.format(
dag_id, task_id, date.isoformat(), self.ti.try_number
).replace(':', '_')
moto.core.moto_api_backend.reset()
self.conn = boto3.client('logs', region_name=self.region_name)
def test_hook_raises(self):
handler = CloudwatchTaskHandler(
self.local_log_location,
f"arn:aws:logs:{self.region_name}:11111111:log-group:{self.remote_log_group}",
self.filename_template,
)
with mock.patch.object(handler.log, 'error') as mock_error:
with mock.patch("airflow.providers.amazon.aws.hooks.logs.AwsLogsHook") as mock_hook:
mock_hook.side_effect = Exception('Failed to connect')
# Initialize the hook
handler.hook
mock_error.assert_called_once_with(
'Could not create an AwsLogsHook with connection id "%s". Please make '
'sure that airflow[aws] is installed and the Cloudwatch logs connection exists.',
'aws_default',
)
class TestCloudwatchTaskHandler(unittest.TestCase):
@conf_vars({('logging', 'remote_log_conn_id'): 'aws_default'})
def setUp(self):
self.remote_log_group = 'log_group_name'
self.region_name = 'us-west-2'
self.local_log_location = 'local/log/location'
self.filename_template = '{dag_id}/{task_id}/{execution_date}/{try_number}.log'
self.cloudwatch_task_handler = CloudwatchTaskHandler(
self.local_log_location,
f"arn:aws:logs:{self.region_name}:11111111:log-group:{self.remote_log_group}",
self.filename_template,
)
self.cloudwatch_task_handler.hook
date = datetime(2020, 1, 1)
dag_id = 'dag_for_testing_file_task_handler'
task_id = 'task_for_testing_file_log_handler'
self.dag = DAG(dag_id=dag_id, start_date=date)
task = DummyOperator(task_id=task_id, dag=self.dag)
self.ti = TaskInstance(task=task, execution_date=date)
self.ti.try_number = 1
self.ti.state = State.RUNNING
self.remote_log_stream = '{}/{}/{}/{}.log'.format(
dag_id, task_id, date.isoformat(), self.ti.try_number
).replace(':', '_')
moto.core.moto_api_backend.reset()
self.conn = boto3.client('logs', region_name=self.region_name)
def tearDown(self):
self.cloudwatch_task_handler.handler = None
def test_hook(self):
assert isinstance(self.cloudwatch_task_handler.hook, AwsLogsHook)
@conf_vars({('logging', 'remote_log_conn_id'): 'aws_default'})
def test_hook_raises(self):
handler = CloudwatchTaskHandler(
self.local_log_location,
f"arn:aws:logs:{self.region_name}:11111111:log-group:{self.remote_log_group}",
self.filename_template,
)
with mock.patch.object(handler.log, 'error') as mock_error:
with mock.patch("airflow.providers.amazon.aws.hooks.logs.AwsLogsHook") as mock_hook:
mock_hook.side_effect = Exception('Failed to connect')
# Initialize the hook
handler.hook
mock_error.assert_called_once_with(
'Could not create an AwsLogsHook with connection id "%s". Please make '
'sure that apache-airflow[aws] is installed and the Cloudwatch '
'logs connection exists. Exception: "%s"',
'aws_default',
ANY,
)
def test_handler(self):
self.cloudwatch_task_handler.set_context(self.ti)
assert isinstance(self.cloudwatch_task_handler.handler, CloudWatchLogHandler)
def test_write(self):
handler = self.cloudwatch_task_handler
handler.set_context(self.ti)
messages = [str(i) for i in range(10)]
with mock.patch("watchtower.CloudWatchLogHandler.emit") as mock_emit:
for message in messages:
handler.handle(message)
mock_emit.assert_has_calls([call(message) for message in messages])
def test_event_to_str(self):
handler = self.cloudwatch_task_handler
events = [
{'timestamp': 1617400267123, 'message': 'First'},
{'timestamp': 1617400367456, 'message': 'Second'},
{'timestamp': 1617400467789, 'message': 'Third'},
]
assert [handler._event_to_str(event) for event in events] == (
[
'[2021-04-02 21:51:07,123] First',
'[2021-04-02 21:52:47,456] Second',
'[2021-04-02 21:54:27,789] Third',
]
)
def test_read(self):
# Confirmed via AWS Support call:
# CloudWatch events must be ordered chronologically otherwise
# boto3 put_log_event API throws InvalidParameterException
# (moto does not throw this exception)
generate_log_events(
self.conn,
self.remote_log_group,
self.remote_log_stream,
[
{'timestamp': 1617400267123, 'message': 'First'},
{'timestamp': 1617400367456, 'message': 'Second'},
{'timestamp': 1617400467789, 'message': 'Third'},
],
)
msg_template = '*** Reading remote log from Cloudwatch log_group: {} log_stream: {}.\n{}\n'
events = '\n'.join(
[
'[2021-04-02 21:51:07,123] First',
'[2021-04-02 21:52:47,456] Second',
'[2021-04-02 21:54:27,789] Third',
]
)
assert self.cloudwatch_task_handler.read(self.ti) == (
[[('', msg_template.format(self.remote_log_group, self.remote_log_stream, events))]],
[{'end_of_log': True}],
)
def test_read_wrong_log_stream(self):
generate_log_events(
self.conn,
self.remote_log_group,
'alternate_log_stream',
[
{'timestamp': 10000, 'message': 'First'},
{'timestamp': 20000, 'message': 'Second'},
{'timestamp': 30000, 'message': 'Third'},
],
)
msg_template = '*** Reading remote log from Cloudwatch log_group: {} log_stream: {}.\n{}\n'
error_msg = 'Could not read remote logs from log_group: {} log_stream: {}.'.format(
self.remote_log_group, self.remote_log_stream
)
assert self.cloudwatch_task_handler.read(self.ti) == (
[[('', msg_template.format(self.remote_log_group, self.remote_log_stream, error_msg))]],
[{'end_of_log': True}],
)
def test_read_wrong_log_group(self):
generate_log_events(
self.conn,
'alternate_log_group',
self.remote_log_stream,
[
{'timestamp': 10000, 'message': 'First'},
{'timestamp': 20000, 'message': 'Second'},
{'timestamp': 30000, 'message': 'Third'},
],
)
msg_template = '*** Reading remote log from Cloudwatch log_group: {} log_stream: {}.\n{}\n'
error_msg = 'Could not read remote logs from log_group: {} log_stream: {}.'.format(
self.remote_log_group, self.remote_log_stream
)
assert self.cloudwatch_task_handler.read(self.ti) == (
[[('', msg_template.format(self.remote_log_group, self.remote_log_stream, error_msg))]],
[{'end_of_log': True}],
)
def test_close_prevents_duplicate_calls(self):
with mock.patch("watchtower.CloudWatchLogHandler.close") as mock_log_handler_close:
with mock.patch("airflow.utils.log.file_task_handler.FileTaskHandler.set_context"):
self.cloudwatch_task_handler.set_context(self.ti)
for _ in range(5):
self.cloudwatch_task_handler.close()
mock_log_handler_close.assert_called_once()
class TestCloudwatchTaskHandler(unittest.TestCase):
@conf_vars({('logging', 'remote_log_conn_id'): 'aws_default'})
def setUp(self):
self.remote_log_group = 'log_group_name'
self.region_name = 'us-west-2'
self.local_log_location = 'local/log/location'
self.filename_template = '{dag_id}/{task_id}/{execution_date}/{try_number}.log'
self.cloudwatch_task_handler = CloudwatchTaskHandler(
self.local_log_location,
f"arn:aws:logs:{self.region_name}:11111111:log-group:{self.remote_log_group}",
self.filename_template,
)
self.cloudwatch_task_handler.hook
date = datetime(2020, 1, 1)
dag_id = 'dag_for_testing_file_task_handler'
task_id = 'task_for_testing_file_log_handler'
self.dag = DAG(dag_id=dag_id, start_date=date)
task = DummyOperator(task_id=task_id, dag=self.dag)
self.ti = TaskInstance(task=task, execution_date=date)
self.ti.try_number = 1
self.ti.state = State.RUNNING
self.remote_log_stream = '{}/{}/{}/{}.log'.format(
dag_id, task_id, date.isoformat(),
self.ti.try_number).replace(':', '_')
moto.core.moto_api_backend.reset()
self.conn = boto3.client('logs', region_name=self.region_name)
def tearDown(self):
self.cloudwatch_task_handler.handler = None
def test_hook(self):
self.assertIsInstance(self.cloudwatch_task_handler.hook, AwsLogsHook)
@conf_vars({('logging', 'remote_log_conn_id'): 'aws_default'})
def test_hook_raises(self):
handler = CloudwatchTaskHandler(
self.local_log_location,
f"arn:aws:logs:{self.region_name}:11111111:log-group:{self.remote_log_group}",
self.filename_template,
)
with mock.patch.object(handler.log, 'error') as mock_error:
with mock.patch(
"airflow.providers.amazon.aws.hooks.logs.AwsLogsHook"
) as mock_hook:
mock_hook.side_effect = Exception('Failed to connect')
# Initialize the hook
handler.hook
mock_error.assert_called_once_with(
'Could not create an AwsLogsHook with connection id "%s". Please make '
'sure that airflow[aws] is installed and the Cloudwatch logs connection exists.',
'aws_default',
)
def test_handler(self):
self.cloudwatch_task_handler.set_context(self.ti)
self.assertIsInstance(self.cloudwatch_task_handler.handler,
CloudWatchLogHandler)
def test_write(self):
handler = self.cloudwatch_task_handler
handler.set_context(self.ti)
messages = [str(i) for i in range(10)]
with mock.patch("watchtower.CloudWatchLogHandler.emit") as mock_emit:
for message in messages:
handler.handle(message)
mock_emit.assert_has_calls([call(message) for message in messages])
def test_read(self):
generate_log_events(
self.conn,
self.remote_log_group,
self.remote_log_stream,
[
{
'timestamp': 20000,
'message': 'Second'
},
{
'timestamp': 10000,
'message': 'First'
},
{
'timestamp': 30000,
'message': 'Third'
},
],
)
expected = (
'*** Reading remote log from Cloudwatch log_group: {} log_stream: {}.\nFirst\nSecond\nThird\n'
)
self.assertEqual(
self.cloudwatch_task_handler.read(self.ti),
(
[[('',
expected.format(self.remote_log_group,
self.remote_log_stream))]],
[{
'end_of_log': True
}],
),
)
def test_read_wrong_log_stream(self):
generate_log_events(
self.conn,
self.remote_log_group,
'alternate_log_stream',
[
{
'timestamp': 20000,
'message': 'Second'
},
{
'timestamp': 10000,
'message': 'First'
},
{
'timestamp': 30000,
'message': 'Third'
},
],
)
msg_template = '*** Reading remote log from Cloudwatch log_group: {} log_stream: {}.\n{}\n'
error_msg = 'Could not read remote logs from log_group: {} log_stream: {}.'.format(
self.remote_log_group, self.remote_log_stream)
self.assertEqual(
self.cloudwatch_task_handler.read(self.ti),
(
[[('',
msg_template.format(self.remote_log_group,
self.remote_log_stream, error_msg))]],
[{
'end_of_log': True
}],
),
)
def test_read_wrong_log_group(self):
generate_log_events(
self.conn,
'alternate_log_group',
self.remote_log_stream,
[
{
'timestamp': 20000,
'message': 'Second'
},
{
'timestamp': 10000,
'message': 'First'
},
{
'timestamp': 30000,
'message': 'Third'
},
],
)
msg_template = '*** Reading remote log from Cloudwatch log_group: {} log_stream: {}.\n{}\n'
error_msg = 'Could not read remote logs from log_group: {} log_stream: {}.'.format(
self.remote_log_group, self.remote_log_stream)
self.assertEqual(
self.cloudwatch_task_handler.read(self.ti),
(
[[('',
msg_template.format(self.remote_log_group,
self.remote_log_stream, error_msg))]],
[{
'end_of_log': True
}],
),
)
def test_close_prevents_duplicate_calls(self):
with mock.patch("watchtower.CloudWatchLogHandler.close"
) as mock_log_handler_close:
with mock.patch(
"airflow.utils.log.file_task_handler.FileTaskHandler.set_context"
):
self.cloudwatch_task_handler.set_context(self.ti)
for _ in range(5):
self.cloudwatch_task_handler.close()
mock_log_handler_close.assert_called_once()