def download_s3_logs(args):
sys.stderr.write(colored('Checking for S3 log files', 'cyan') + '\n')
logs = logs_for_all_requests(args)
async_requests = []
zipped_files = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(args, time_from_filename(filename)):
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest('GET', log_file['getUrl'], callback=generate_callback(log_file['getUrl'], args.dest, filename, args.chunk_size, args.verbose))
)
else:
if args.verbose:
sys.stderr.write(colored('Log already downloaded {0}'.format(filename), 'magenta') + '\n')
all_logs.append('{0}/{1}'.format(args.dest, filename.replace('.gz', '.log')))
zipped_files.append('{0}/{1}'.format(args.dest, filename))
else:
if args.verbose:
sys.stderr.write(colored('Excluding {0}, not in date range'.format(filename), 'magenta') + '\n')
if async_requests:
sys.stderr.write(colored('Starting S3 Downloads with {0} parallel fetches'.format(args.num_parallel_fetches), 'cyan'))
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
else:
sys.stderr.write(colored('No S3 logs to download', 'cyan'))
sys.stderr.write(colored('\nUnpacking S3 logs\n', 'cyan'))
all_logs = all_logs + logfetch_base.unpack_logs(args, zipped_files)
sys.stderr.write(colored('All S3 logs up to date', 'cyan') + '\n')
return all_logs
def download_live_logs(args):
tasks = tasks_to_check(args)
async_requests = []
zipped_files = []
all_logs = []
sys.stderr.write(colored('Finding current live log files', 'cyan') + '\n')
for task in tasks:
metadata = files_json(args, task)
if 'slaveHostname' in metadata:
uri = DOWNLOAD_FILE_FORMAT.format(metadata['slaveHostname'])
if args.verbose:
sys.stderr.write(colored('Finding logs in base directory on {0}'.format(metadata['slaveHostname']), 'magenta') + '\n')
for log_file in base_directory_files(args, task, metadata):
logfile_name = '{0}-{1}'.format(task, log_file)
if not args.logtype or (args.logtype and logfetch_base.log_matches(log_file, args.logtype.replace('logs/', ''))):
if should_download(args, logfile_name, task):
async_requests.append(
grequests.AsyncRequest('GET',uri ,
callback=generate_callback(uri, args.dest, logfile_name, args.chunk_size, args.verbose),
params={'path' : '{0}/{1}/{2}'.format(metadata['fullPathToRoot'], metadata['currentDirectory'], log_file)},
headers=args.headers
)
)
if logfile_name.endswith('.gz'):
zipped_files.append('{0}/{1}'.format(args.dest, logfile_name))
else:
all_logs.append('{0}/{1}'.format(args.dest, logfile_name.replace('.gz', '.log')))
elif args.logtype and args.verbose:
sys.stderr.write(colored('Excluding log {0}, doesn\'t match {1}'.format(log_file, args.logtype), 'magenta') + '\n')
if args.verbose:
sys.stderr.write(colored('Finding logs in logs directory on {0}'.format(metadata['slaveHostname']), 'magenta') + '\n')
for log_file in logs_folder_files(args, task):
logfile_name = '{0}-{1}'.format(task, log_file)
if not args.logtype or (args.logtype and logfetch_base.log_matches(log_file, args.logtype.replace('logs/', ''))):
if should_download(args, logfile_name, task):
async_requests.append(
grequests.AsyncRequest('GET',uri ,
callback=generate_callback(uri, args.dest, logfile_name, args.chunk_size, args.verbose),
params={'path' : '{0}/{1}/logs/{2}'.format(metadata['fullPathToRoot'], metadata['currentDirectory'], log_file)},
headers=args.headers
)
)
if logfile_name.endswith('.gz'):
zipped_files.append('{0}/{1}'.format(args.dest, logfile_name))
else:
all_logs.append('{0}/{1}'.format(args.dest, logfile_name.replace('.gz', '.log')))
elif args.logtype and args.verbose:
sys.stderr.write(colored('Excluding log {0}, doesn\'t match {1}'.format(log_file, args.logtype), 'magenta') + '\n')
if async_requests:
sys.stderr.write(colored('Starting live logs downloads\n', 'cyan'))
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
if zipped_files:
sys.stderr.write(colored('\nUnpacking logs\n', 'cyan'))
all_logs = all_logs + logfetch_base.unpack_logs(args, zipped_files)
return all_logs
def download_s3_logs(args):
sys.stderr.write(colored('Checking for S3 log files', 'cyan') + '\n')
logs = logs_for_all_requests(args)
async_requests = []
zipped_files = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(
args, int(str(log_file['lastModified'])[0:-3])):
if not args.logtype or log_matches(args, filename):
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest('GET',
log_file['getUrl'],
callback=generate_callback(
log_file['getUrl'],
args.dest, filename,
args.chunk_size,
args.verbose),
headers=args.headers))
else:
if args.verbose:
sys.stderr.write(
colored(
'Log already downloaded {0}'.format(filename),
'magenta') + '\n')
all_logs.append('{0}/{1}'.format(
args.dest, filename.replace('.gz', '.log')))
zipped_files.append('{0}/{1}'.format(args.dest, filename))
else:
if args.verbose:
sys.stderr.write(
colored(
'Excluding {0} log does not match logtype argument {1}'
.format(filename, args.logtype), 'magenta') + '\n')
else:
if args.verbose:
sys.stderr.write(
colored(
'Excluding {0}, not in date range'.format(filename),
'magenta') + '\n')
if async_requests:
sys.stderr.write(
colored(
'Starting S3 Downloads with {0} parallel fetches'.format(
args.num_parallel_fetches), 'cyan'))
grequests.map(async_requests,
stream=True,
size=args.num_parallel_fetches)
else:
sys.stderr.write(colored('No S3 logs to download', 'cyan'))
sys.stderr.write(colored('\nUnpacking S3 logs\n', 'cyan'))
all_logs = all_logs + logfetch_base.unpack_logs(args, zipped_files)
sys.stderr.write(colored('All S3 logs up to date', 'cyan') + '\n')
return all_logs
def download_s3_logs(args):
sys.stderr.write(colored('Checking for S3 log files', 'cyan') + '\n')
logs = logs_for_all_requests(args)
async_requests = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(args, time_from_filename(filename)):
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest('GET', log_file['getUrl'], callback=generate_callback(log_file['getUrl'], args.dest, filename, args.chunk_size))
)
all_logs.append('{0}/{1}'.format(args.dest, filename.replace('.gz', '.log')))
if async_requests:
sys.stderr.write(colored('Starting S3 Downloads', 'cyan'))
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
zipped_files = ['{0}/{1}'.format(args.dest, log_file['key'].rsplit("/", 1)[1]) for log_file in logs]
sys.stderr.write(colored('Unpacking S3 logs\n', 'cyan'))
logfetch_base.unpack_logs(zipped_files)
sys.stderr.write(colored('All S3 logs up to date', 'cyan') + '\n')
return all_logs
def download_s3_logs(args):
sys.stderr.write(colored('Checking for S3 log files', 'blue') + '\n')
logs = get_json_response(singularity_s3logs_uri(args))
async_requests = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
full_log_path = '{0}/{1}'.format(args.dest, filename.replace('.gz', '.log'))
full_gz_path = '{0}/{1}'.format(args.dest, filename)
if in_date_range(args, filename):
if not (os.path.isfile(full_log_path) or os.path.isfile(full_gz_path)):
async_requests.append(
grequests.AsyncRequest('GET', log_file['getUrl'],
callback=generate_callback(log_file['getUrl'], args.dest, filename, args.chunk_size)
)
)
all_logs.append('{0}/{1}'.format(args.dest, filename.replace('.gz', '.log')))
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
zipped_files = ['{0}/{1}'.format(args.dest, log_file['key'].rsplit("/", 1)[1]) for log_file in logs]
logfetch_base.unpack_logs(zipped_files)
sys.stderr.write(colored('All S3 logs up to date', 'blue') + '\n')
return all_logs
def download_s3_logs(args):
if not args.silent:
sys.stderr.write(colored('Checking for S3 log files', 'cyan') + '\n')
callbacks.progress = 0
logs = logs_for_all_requests(args)
async_requests = []
zipped_files = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(args, int(str(log_file['lastModified'])[0:-3])):
if not args.logtype or log_matches(args, filename):
if args.verbose and not args.silent:
sys.stderr.write(colored('Including log {0}'.format(filename), 'blue') + '\n')
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest('GET', log_file['getUrl'], callback=callbacks.generate_callback(log_file['getUrl'], args.dest, filename, args.chunk_size, args.verbose, args.silent), headers=args.headers)
)
else:
if args.verbose and not args.silent:
sys.stderr.write(colored('Log already downloaded {0}'.format(filename), 'blue') + '\n')
all_logs.append('{0}/{1}'.format(args.dest, filename.replace('.gz', '.log')))
zipped_files.append('{0}/{1}'.format(args.dest, filename))
else:
if args.verbose and not args.silent:
sys.stderr.write(colored('Excluding {0} log does not match logtype argument {1}'.format(filename, args.logtype), 'magenta') + '\n')
else:
if args.verbose and not args.silent:
sys.stderr.write(colored('Excluding {0}, not in date range'.format(filename), 'magenta') + '\n')
if async_requests:
if not args.silent:
sys.stderr.write(colored('Starting {0} S3 Downloads with {1} parallel fetches\n'.format(len(async_requests), args.num_parallel_fetches), 'cyan'))
callbacks.goal = len(async_requests)
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
if not args.silent and not args.download_only:
sys.stderr.write(colored('\nUnpacking {0} S3 log(s)\n'.format(len(async_requests)), 'cyan'))
else:
if not args.silent:
sys.stderr.write(colored('No S3 logs to download\n', 'cyan'))
if not args.download_only:
all_logs = all_logs + logfetch_base.unpack_logs(args, zipped_files)
if not args.silent:
sys.stderr.write(colored('All S3 logs up to date\n', 'cyan'))
return all_logs
def download_live_logs(args):
tasks = tasks_to_check(args)
async_requests = []
zipped_files = []
all_logs = []
sys.stderr.write(colored('Finding current live log files', 'cyan') + '\n')
for task in tasks:
metadata = files_json(args, task)
if 'slaveHostname' in metadata:
uri = DOWNLOAD_FILE_FORMAT.format(metadata['slaveHostname'])
if args.verbose:
sys.stderr.write(
colored(
'Finding logs in base directory on {0}'.format(
metadata['slaveHostname']), 'magenta') + '\n')
for log_file in base_directory_files(args, task, metadata):
logfile_name = '{0}-{1}'.format(task, log_file)
if not args.logtype or (
args.logtype and logfetch_base.log_matches(
log_file, args.logtype.replace('logs/', ''))):
if should_download(args, logfile_name, task):
async_requests.append(
grequests.AsyncRequest(
'GET',
uri,
callback=generate_callback(
uri, args.dest, logfile_name,
args.chunk_size, args.verbose),
params={
'path':
'{0}/{1}/{2}'.format(
metadata['fullPathToRoot'],
metadata['currentDirectory'], log_file)
},
headers=args.headers))
if logfile_name.endswith('.gz'):
zipped_files.append('{0}/{1}'.format(
args.dest, logfile_name))
else:
all_logs.append('{0}/{1}'.format(
args.dest, logfile_name.replace('.gz', '.log')))
elif args.logtype and args.verbose:
sys.stderr.write(
colored(
'Excluding log {0}, doesn\'t match {1}'.format(
log_file, args.logtype), 'magenta') + '\n')
if args.verbose:
sys.stderr.write(
colored(
'Finding logs in logs directory on {0}'.format(
metadata['slaveHostname']), 'magenta') + '\n')
for log_file in logs_folder_files(args, task):
logfile_name = '{0}-{1}'.format(task, log_file)
if not args.logtype or (
args.logtype and logfetch_base.log_matches(
log_file, args.logtype.replace('logs/', ''))):
if should_download(args, logfile_name, task):
async_requests.append(
grequests.AsyncRequest(
'GET',
uri,
callback=generate_callback(
uri, args.dest, logfile_name,
args.chunk_size, args.verbose),
params={
'path':
'{0}/{1}/logs/{2}'.format(
metadata['fullPathToRoot'],
metadata['currentDirectory'], log_file)
},
headers=args.headers))
if logfile_name.endswith('.gz'):
zipped_files.append('{0}/{1}'.format(
args.dest, logfile_name))
else:
all_logs.append('{0}/{1}'.format(
args.dest, logfile_name.replace('.gz', '.log')))
elif args.logtype and args.verbose:
sys.stderr.write(
colored(
'Excluding log {0}, doesn\'t match {1}'.format(
log_file, args.logtype), 'magenta') + '\n')
if async_requests:
sys.stderr.write(colored('Starting live logs downloads\n', 'cyan'))
grequests.map(async_requests,
stream=True,
size=args.num_parallel_fetches)
if zipped_files:
sys.stderr.write(colored('\nUnpacking logs\n', 'cyan'))
all_logs = all_logs + logfetch_base.unpack_logs(args, zipped_files)
return all_logs