def download_file(self, file_name_with_path):
"""
Downloads a file from HDFS location identified by the path
Keyword arguments:
file_name_with_path {str} -- Name of the file identified with a path
Returns:
response -- Default Flask response object with file content and appropriate headers set
"""
file_name_with_path = self._get_actual_download_file_path(
file_name_with_path)
open_file_url = self.url + file_name_with_path + "?op=OPEN"
response = RestUtil.request_with_retry().get(open_file_url,
auth=self.auth,
allow_redirects=False)
if response.status_code != 307:
if response.status_code == 404:
raise ObjectNotFoundError(
"File {} not found.".format(file_name_with_path))
raise ServiceError(
"Attempt to open file {0} failed with {1} and {2}.".format(
file_name_with_path, response.status_code,
response.reason))
file_download_url = None
if response.headers is not None:
file_download_url = response.headers["Location"]
if file_download_url is not None:
res = RestUtil.request_with_retry().get(file_download_url,
auth=self.auth,
stream=True)
if not response.ok:
raise ServiceError(
"Attempt to download file {0} failed with {1} and {2}.".
format(file_name_with_path, response.status_code,
response.reason))
response = Response(res.content, headers=dict(res.headers))
response.headers['Content-Type'] = 'application/octet-stream'
response.headers[
'Content-Disposition'] = 'attachment;filename="{}"'.format(
file_name_with_path.split("/")[-1])
return response
def get_job_status(self, job_id):
"""
Fetches the status of the batch job using Livy's batches endpoint
Keyword arguments:
job_id {str} -- Job identifier
Returns:
response {dict} -- Dictionary with job id, state and the application id.
"""
job_url = "{}/batches/{}".format(self.url, job_id)
response = RestUtil.request_with_retry().get(url=job_url,
auth=self.auth)
if not response.ok:
if response.status_code == 404:
raise ObjectNotFoundError(
"Job with id {} not found.".format(job_id))
raise ServiceError("Failed to get jobs state. " + response.text)
job_response = response.json()
response = {
"id": job_response.get("id"),
"state": job_response.get("state"),
"appId": job_response.get("appId")
}
return response
def get_job_logs(self, job_id, size):
"""
Fetches the logs of the batch job using Livy's batches logs endpoint
Keyword arguments:
job_id {str} -- Job identifier
size {int} -- Number of log lines to be returned
Returns:
response -- Http method response
"""
job_logs_url = "{}/batches/{}/logs".format(self.url, job_id)
if size is not None and size > 0:
job_logs_url = job_logs_url + "?size={}".format(size)
response = RestUtil.request_with_retry().get(url=job_logs_url,
auth=self.auth)
if not response.ok:
if response.status_code == 404:
raise ObjectNotFoundError(
"Job with id {} not found.".format(job_id))
raise ServiceError("Failed to get job logs. " + response.text)
return response
def delete_file(self, file_name_with_path):
"""
Deletes a file from HDFS identified by the path
Keyword arguments:
file_name_with_path {str} -- Name of the file identified with a path
Returns:
response -- Http method response
"""
delete_file_url = self.url + file_name_with_path + "?op=DELETE"
if os.path.splitext(file_name_with_path)[-1] == "":
delete_file_url = delete_file_url + "&recursive=true"
response = RestUtil.request_with_retry().delete(delete_file_url,
auth=self.auth)
if not response.ok:
raise ServiceError(
"Attempt to delete file {0} failed with {1} and {2}.".format(
file_name_with_path, response.status_code,
response.reason))
return response
def run_batch_job(self,
job_json,
background=True,
timeout=constants.SYNC_JOB_MAX_WAIT_TIME):
"""
Submits a job request using Livy batches endpoint
Keyword arguments:
job_json {str} -- Job request payload
background {bool} -- Flag indicating if the method should wait until the job finishes or return immediately after submitting the request.
Returns:
response {dict} -- Dictionary with job id, state and the application id.
"""
job_url = "{}/batches".format(self.url)
response = RestUtil.request_with_retry().post(
url=job_url,
json=job_json,
headers={"Content-Type": "application/json"},
auth=self.auth)
if not response.ok:
raise ServiceError("Failed to run job. " + response.text)
job_response = response.json()
job_id = job_response.get("id")
state = job_response.get("state")
appId = job_response.get("appId")
if background is False:
start_time = time.time()
elapsed_time = 0
sleep_time = 15
while state not in (constants.LIVY_JOB_FINISHED_STATE,
constants.LIVY_JOB_FAILED_STATE,
constants.LIVY_JOB_DEAD_STATE,
constants.LIVY_JOB_KILLED_STATE):
if elapsed_time > int(timeout):
raise ServiceError(
"Job didn't come to Finished/Failed state in {} seconds. Current state is "
.format(timeout, state))
print("{}: Sleeping for {} seconds. Current state {}".format(
datetime.datetime.now(), sleep_time, state))
sleep(sleep_time)
elapsed_time = time.time() - start_time
status = self.get_job_status(job_id)
state = status.get("state")
appId = status.get("appId")
response = {"id": job_id, "state": state, "appId": appId}
return response
def _get_actual_download_file_path(self, file_name_with_path):
download_file_path = None
list_status_url = self.url + file_name_with_path + "?op=LISTSTATUS"
response = RestUtil.request_with_retry().get(list_status_url,
auth=self.auth)
if not response.ok:
if response.status_code == 404:
raise ObjectNotFoundError(
"File {} not found.".format(file_name_with_path))
raise ServiceError(
"Attempt to open file {0} failed with {1} and {2}.".format(
file_name_with_path, response.status_code,
response.reason))
list_status_response = json.loads(response.text)
if list_status_response is not None and list_status_response.get(
"FileStatuses") is not None:
files_statuses = list_status_response.get("FileStatuses")
if files_statuses.get("FileStatus") is not None:
file_status_list = files_statuses.get("FileStatus")
if len(file_status_list) > 1:
raise BadRequestError(
"Specified path is a directory containing multiple files. Supported only if single part file is inside folder."
)
path_suffix = file_status_list[0]["pathSuffix"]
if len(path_suffix) > 0:
if file_status_list[0]["type"] == "DIRECTORY":
download_file_path = self._get_actual_download_file_path(
file_name_with_path + "/" + path_suffix)
elif file_status_list[0]["type"] == "FILE":
download_file_path = file_name_with_path + "/" + path_suffix
else:
download_file_path = file_name_with_path
return download_file_path
def upload_file(self, file_name_with_path, data, overwrite=False):
"""
Uploads file to a HDFS location identified by the path
Keyword arguments:
file_name_with_path {str} -- Name of the file identified with a path
data {bytearray} -- Byte array representation of the file
overwrite {bool} -- Flag indicating of the file should be overwritten
Returns:
response {dict} -- Dictionary denoting the status of the upload operation and the relative location of the file.
"""
# The following lines (114-123) are a temporary fix until we have the right change made in the
# ibm-wos-utils module to handle upload of this specific file only when its not found
# in the HDFS location
if "main_job.py" in file_name_with_path:
check_file_status_url = self.url + file_name_with_path + "?op=LISTSTATUS"
response = RestUtil.request_with_retry().get(check_file_status_url,
auth=self.auth)
if response.status_code == 200:
logger.log_warning("File already exists.. Skipping upload...")
response = {
"status": "finished",
"location": file_name_with_path
}
return response
create_file_url = self.url + file_name_with_path + "?op=CREATE"
if overwrite:
create_file_url = create_file_url + "&overwrite=true"
response = RestUtil.request_with_retry().put(create_file_url,
auth=self.auth,
allow_redirects=False)
if response.status_code != 307:
raise ServiceError(
"Attempt to create file {0} failed with {1} and {2}.".format(
file_name_with_path, response.status_code,
response.reason))
file_write_url = None
if response.headers is not None:
file_write_url = response.headers["Location"]
if file_write_url is not None:
response = RestUtil.request_with_retry().put(file_write_url,
auth=self.auth,
data=data)
retry_attempt = 0
sleep_factor = random.randint(1, 5)
# If the file upload fails with 404, during multiple parallel requests trying to upload the same file,
# attempting retry up-to 5 times with a random start sleep time ranging between 1 and 5 seconds
# and a back-off factor of 1.5
if response.status_code == 404:
while retry_attempt < 5:
sleep_factor = sleep_factor * 1.5
time.sleep(sleep_factor)
retry_attempt += 1
logger.log_info("Re-attempt {} of file {} upload.".format(
retry_attempt, file_name_with_path))
actual_response = RestUtil.request_with_retry().put(
create_file_url, auth=self.auth, allow_redirects=False)
if actual_response.headers is not None:
file_url = actual_response.headers["Location"]
response = RestUtil.request_with_retry().put(
file_url, auth=self.auth, data=data)
if not response.ok:
raise ServiceError(
"Attempt to write to file {0} failed with {1} and {2}.".
format(file_name_with_path, response.status_code,
response.reason))
check_file_status_url = self.url + file_name_with_path + "?op=LISTSTTAUS"
response = RestUtil.request_with_retry().get(check_file_status_url,
auth=self.auth)
if response.status_code == 404:
raise ServiceError(
"File {} not found".format(file_name_with_path))
response = {"status": "finished", "location": file_name_with_path}
return response