def test_read_log_stream_should_read_each_try_in_turn(self, mock_read):
first_return = ([[('', "try_number=1.")]], [{"end_of_log": True}])
second_return = ([[('', "try_number=2.")]], [{"end_of_log": True}])
third_return = ([[('', "try_number=3.")]], [{"end_of_log": True}])
fourth_return = ([[('', "should never be read")]], [{
"end_of_log": True
}])
mock_read.side_effect = [
first_return, second_return, third_return, fourth_return
]
task_log_reader = TaskLogReader()
log_stream = task_log_reader.read_log_stream(ti=self.ti,
try_number=None,
metadata={})
self.assertEqual(
['\ntry_number=1.\n', '\ntry_number=2.\n', '\ntry_number=3.\n'],
list(log_stream))
mock_read.assert_has_calls(
[
mock.call(self.ti, 1, metadata={}),
mock.call(self.ti, 2, metadata={}),
mock.call(self.ti, 3, metadata={}),
],
any_order=False,
)
def test_read_log_stream_should_support_multiple_chunks(self, mock_read):
first_return = ([[('', "1st line")]], [{}])
second_return = ([[('', "2nd line")]], [{"end_of_log": False}])
third_return = ([[('', "3rd line")]], [{"end_of_log": True}])
fourth_return = ([[('', "should never be read")]], [{
"end_of_log": True
}])
mock_read.side_effect = [
first_return, second_return, third_return, fourth_return
]
task_log_reader = TaskLogReader()
log_stream = task_log_reader.read_log_stream(ti=self.ti,
try_number=1,
metadata={})
self.assertEqual(["\n1st line\n", "\n2nd line\n", "\n3rd line\n"],
list(log_stream))
mock_read.assert_has_calls(
[
mock.call(self.ti, 1, metadata={}),
mock.call(self.ti, 1, metadata={}),
mock.call(self.ti, 1, metadata={"end_of_log": False}),
],
any_order=False,
)
def get_log(session, dag_id, dag_run_id, task_id, task_try_number, full_content=False, token=None):
"""Get logs for specific task instance"""
key = current_app.config["SECRET_KEY"]
if not token:
metadata = {}
else:
try:
metadata = URLSafeSerializer(key).loads(token)
except BadSignature:
raise BadRequest("Bad Signature. Please use only the tokens provided by the API.")
if metadata.get('download_logs') and metadata['download_logs']:
full_content = True
if full_content:
metadata['download_logs'] = True
else:
metadata['download_logs'] = False
task_log_reader = TaskLogReader()
if not task_log_reader.supports_read:
raise BadRequest("Task log handler does not support read logs.")
ti = (
session.query(TaskInstance)
.filter(TaskInstance.task_id == task_id, TaskInstance.run_id == dag_run_id)
.join(TaskInstance.dag_run)
.options(eagerload(TaskInstance.dag_run))
.one_or_none()
)
if ti is None:
metadata['end_of_log'] = True
raise NotFound(title="TaskInstance not found")
dag = current_app.dag_bag.get_dag(dag_id)
if dag:
try:
ti.task = dag.get_task(ti.task_id)
except TaskNotFound:
pass
return_type = request.accept_mimetypes.best_match(['text/plain', 'application/json'])
# return_type would be either the above two or None
if return_type == 'application/json' or return_type is None: # default
logs, metadata = task_log_reader.read_log_chunks(ti, task_try_number, metadata)
logs = logs[0] if task_try_number is not None else logs
token = URLSafeSerializer(key).dumps(metadata)
return logs_schema.dump(LogResponseObject(continuation_token=token, content=logs))
# text/plain. Stream
logs = task_log_reader.read_log_stream(ti, task_try_number, metadata)
return Response(logs, headers={"Content-Type": return_type})
def test_test_test_read_log_stream_should_read_one_try(self):
task_log_reader = TaskLogReader()
stream = task_log_reader.read_log_stream(ti=self.ti,
try_number=1,
metadata={})
assert [
"\n*** Reading local file: "
f"{self.log_dir}/dag_log_reader/task_log_reader/2017-09-01T00.00.00+00.00/1.log\n"
"try_number=1.\n"
"\n"
] == list(stream)
def assert_remote_logs(self, expected_message, ti):
with provide_gcp_context(GCP_STACKDDRIVER), conf_vars({
('logging', 'remote_logging'): 'True',
('logging', 'remote_base_log_folder'): f"stackdriver://{self.log_name}",
}):
from airflow.config_templates import airflow_local_settings
importlib.reload(airflow_local_settings)
settings.configure_logging()
task_log_reader = TaskLogReader()
logs = "\n".join(task_log_reader.read_log_stream(ti, try_number=None, metadata={}))
self.assertIn(expected_message, logs)
def assert_remote_logs(self, expected_message, ti):
with provide_gcp_context(GCP_GCS_KEY), conf_vars({
('logging', 'remote_logging'): 'True',
('logging', 'remote_base_log_folder'): f"gs://{self.bucket_name}/path/to/logs",
('logging', 'remote_log_conn_id'): "google_cloud_default",
}):
from airflow.config_templates import airflow_local_settings
importlib.reload(airflow_local_settings)
settings.configure_logging()
task_log_reader = TaskLogReader()
logs = "\n".join(task_log_reader.read_log_stream(ti, try_number=None, metadata={}))
self.assertIn(expected_message, logs)
def test_test_test_read_log_stream_should_read_all_logs(self):
task_log_reader = TaskLogReader()
stream = task_log_reader.read_log_stream(ti=self.ti, try_number=None, metadata={})
self.assertEqual(
[
"\n*** Reading local file: "
f"{self.log_dir}/dag_log_reader/task_log_reader/2017-09-01T00.00.00+00.00/1.log\n"
"try_number=1.\n"
"\n",
"\n*** Reading local file: "
f"{self.log_dir}/dag_log_reader/task_log_reader/2017-09-01T00.00.00+00.00/2.log\n"
"try_number=2.\n"
"\n",
"\n*** Reading local file: "
f"{self.log_dir}/dag_log_reader/task_log_reader/2017-09-01T00.00.00+00.00/3.log\n"
"try_number=3.\n"
"\n",
],
list(stream),
)
def get_log(
*,
dag_id: str,
dag_run_id: str,
task_id: str,
task_try_number: int,
full_content: bool = False,
map_index: int = -1,
token: Optional[str] = None,
session: Session = NEW_SESSION,
) -> APIResponse:
"""Get logs for specific task instance"""
key = get_airflow_app().config["SECRET_KEY"]
if not token:
metadata = {}
else:
try:
metadata = URLSafeSerializer(key).loads(token)
except BadSignature:
raise BadRequest(
"Bad Signature. Please use only the tokens provided by the API."
)
if metadata.get('download_logs') and metadata['download_logs']:
full_content = True
if full_content:
metadata['download_logs'] = True
else:
metadata['download_logs'] = False
task_log_reader = TaskLogReader()
if not task_log_reader.supports_read:
raise BadRequest("Task log handler does not support read logs.")
ti = (session.query(TaskInstance).filter(
TaskInstance.task_id == task_id,
TaskInstance.dag_id == dag_id,
TaskInstance.run_id == dag_run_id,
TaskInstance.map_index == map_index,
).join(TaskInstance.dag_run).one_or_none())
if ti is None:
metadata['end_of_log'] = True
raise NotFound(title="TaskInstance not found")
dag = get_airflow_app().dag_bag.get_dag(dag_id)
if dag:
try:
ti.task = dag.get_task(ti.task_id)
except TaskNotFound:
pass
return_type = request.accept_mimetypes.best_match(
['text/plain', 'application/json'])
# return_type would be either the above two or None
logs: Any
if return_type == 'application/json' or return_type is None: # default
logs, metadata = task_log_reader.read_log_chunks(
ti, task_try_number, metadata)
logs = logs[0] if task_try_number is not None else logs
# we must have token here, so we can safely ignore it
token = URLSafeSerializer(key).dumps(
metadata) # type: ignore[assignment]
return logs_schema.dump(
LogResponseObject(continuation_token=token, content=logs))
# text/plain. Stream
logs = task_log_reader.read_log_stream(ti, task_try_number, metadata)
return Response(logs, headers={"Content-Type": return_type})